Questions and answers about coronavirus and the UK economy

Risk in the time of Covid-19: what do we know and not know?

Coronavirus has exposed the world’s population to an extreme degree of uncertainty in all dimensions of life. How does this unprecedented global event influence our risk-taking – and how can we measure it reliably?

Our experience of ‘risk’ is changing dramatically. Decisions that we never considered risky before the coronavirus have suddenly entered the forefront of our daily lives: If my mask is in the wash, should I still go to the supermarket? Should I take a crowded bus or pay extra for an Uber? Should I still look to buy a house if I may face redundancy?

The novel conditions under which people are now weighing decisions make predicting behaviour particularly challenging, but we are not starting from scratch. Risk has intrigued economists and psychologists for decades – and we can learn from past crises to help us understand and, in some cases, approximate the impact of choices that people will make today.

Unprecedented times

While Covid-19 has introduced uncertainty in nearly all dimensions of our lives, this is not the first time in human history that we have faced (and adapted to) exceptional circumstances of risk and uncertainty. Economists crucially distinguish between the two: risk is the measure for an event for which we know the probability associated with the outcomes; and uncertainty relates to those events for which we do not know the probabilities of the risky outcomes.

For example, when throwing fair dice, we know the probability of any number coming up is one in six. In the case of Covid-19, in contrast, we do not know the probabilities of the possible outcomes and, as a result, we are living under several sources of major uncertainty. Simply put, we don’t know whether the changes we are experiencing are temporary or an intractable start of a new world.

This is compounded by the fact that we also find ourselves traversing unfamiliar conditions of social isolation and living in fear of an invisible force that may make us sick. Even more pressingly, we are facing these conditions at the same time. What does this mean for decision-making in today’s world?

At this stage, ‘unprecedented’ is the only apparent way to describe the current environment and its effects. But to navigate these compounding factors of uncertainty and risk, we must ask: what other types of risk are similar to those we are experiencing; and does history provide clues to navigating the current global uncertainty?

Many have made comparisons to historical examples of global pandemics, such as the Spanish flu in 1918, which killed approximately 50 million people worldwide. Unfortunately, there are no reliable data for the impact on risk-taking at that time and, given that the Spanish flu occurred at the end of the First World War, today’s choices are wildly different. Where we can draw lessons are the effects of other, more widely understood decisions made under uncertainty. 

Related question: Does the Spanish flu offer lessons in how to tackle a pandemic?

What can research tell us?

A stream of economic research looks at a type of uncertainty known as ‘background risk’ that cannot be avoided, insured against or diversified away (Eeckhoudt et al, 1996; Harrison et al, 2007).

For many, a sudden change in income is a prime example, but Covid-19 also falls into this category. Background uncertainty is out of our immediate control, so it differs from risk experienced in ‘normal’ situations where we have a degree of active choice (‘foreground risk’) such as selecting one option over another in playing a lottery or choosing a financial product.

In the financial domain, these two forms of risk are sometimes also referred to either as endogenous (that is, interactions of options within a set of choices) or exogenous (that is, external shocks: Kimball, 1990; Guiso and Paiella, 2008).

For example, research suggests that financial background uncertainty may result in a reduction in risky asset allocation in order to reduce overall risk exposure (Kimball, 1990). This is in line with what is sometimes called ‘countercyclical risk aversion’ (Guiso et al, 2013; Cohn et al, 2015). Covid-19 has proved devastating to financial markets worldwide and these types of exogenous shocks may currently be at play. 

Related question: What explains stock market reactions to the pandemic?

How does this generalise to non-financial risk-taking?

While background uncertainty is most commonly cited in financial contexts, we may look to understand how background uncertainty can affect other risk-taking domains. More general examples of effects of non-financial exogenous shocks can be found in situations such as natural disasters, wars and terrorist attacks.

Some studies have measured risk-taking after a natural disaster or under exposure to a major life threat by looking at individual choices in lottery games with real monetary wins. Some of the evidence is seemingly counterintuitive. Certain studies suggest an increase in risk-taking after a natural disaster. For example, evacuees from Hurricane Katrina (Eckel et al, 2009), below-poverty-line households in Peru exposed to volcanic threat (Bchir and Willinger, 2013) and flooded households in Pakistan and Australia (Page et al, 2014; Said et al, 2015) all appeared to be more risk-seeking in lottery gambles.

The behavioural mechanism underlying such reactions is unclear and, to the best of our knowledge, has never been fully tested. One potential hypothesis, however, is that the presence of massive uncertainty (the disaster) ‘trumps’ seemingly smaller, more familiar sources of risk (the lottery), which become less salient, less relevant and are perceived as less risky.

Examples of seemingly counterintuitive reactions can be found in non-financial domains: soldiers in a war possibly don’t perceive the risks of smoking, taking drugs or having unprotected sex in the face of great harm or death. Israeli adolescents exposed to continuous threats of terrorist attacks, for example, reported higher levels of risk-taking behaviours such as drinking alcohol or being involved in a fight (Pat-Horenczyk et al, 2007). 

This line of inquiry suggests that greater uncertainty diminishes the perceived smaller risk in comparison. It suggests that risk-taking behaviour may actually increase in the period of and following Covid-19.

Other studies of non-financial exogenous shocks come to conclusions more aligned with exogenous financial risks. Reduced risk-taking in lottery gambles was observed following a tsunami in Thailand (Cassar et al, 2017) and an earthquake or flood in Indonesia (Cameron and Shah, 2015).

This has been interpreted to mean that major background uncertainty induces compensatory behaviour, leading to a reduction in non-essential risk-taking activities, perhaps in an increased effort to protect ourselves. This hypothesis assumes ‘risk homeostasis’ where, being already exposed to the major Covid-19 uncertainty, we reduce our other risk-taking activities, such as smoking or climbing, in order to protect ourselves.

In principle, risk homeostasis or ‘risk compensation’ can also ‘swing’ our behaviour in the opposite direction: for example, taking our multivitamins warrants us having a rich dessert, and wearing a seatbelt means that we can drive faster. More currently, wearing a face mask may make us feel safe to see friends, and without our weekly recreational climb, dark web purchasing may seem all the more attractive. 

Although both these hypotheses are plausible – and the idea of risk compensation has often been proposed in economics and psychology – evidence to date is mixed, to say the least (Houston and Richardson, 2007). Similarly, the ‘large uncertainty trumps small risks’ hypothesis has not yet been systematically validated. This makes offering any more than theorised speculation on the direction of behaviour in times of Covid-19 uncertainty difficult. 

How reliable is the evidence?

Generally, we should not forget that most of the above studies rely on evidence from risk-taking tasks in the form of lottery games with real monetary payments, which have limitations. These games typically have very small stakes, and we know that risk-taking in the face of larger stakes can differ significantly (Holt and Laury, 2002; Fehr-Duda et al, 2010).

Other limitations include the fact that pre-registration and replication is still uncommon in (behavioural and experimental) economics, which makes the reproducibility of these studies in the same or other settings hard to assess. Because systematic research in this area is thus far lacking, systematic reviews or meta-analyses must be conducted to gain a better understanding of these phenomena, and pave a clearer path to predict the impact of current and future shocks to our global risk-taking behaviour. 

Experts are also split on the stability of risk-taking (Frey et al, 2017; Harrison et al, 2019). While both economics and psychology acknowledge the wide heterogeneity in individual risk-taking, there is currently debate on whether risk-taking is a stable trait or one that fluctuates over time and across contexts (Loomes and Pogrebna, 2014; Sanders and Jenkins, 2016).

Evidence is also growing that risk-taking is largely ‘domain-specific’ (Barseghyan et al, 2011; Riddel, 2012; Galizzi et al, 2016). This has a number of plausible implications:

  • If risk-taking is domain-specific, it is difficult to make predictions on the impact on risk-taking ‘in general’. For example, someone who feels at home in the stock market may not be at all comfortable jumping out of a small plane – and vice versa. Similarly, the background uncertainty associated with Covid-19 may influence health domain-specific risk, but may not affect risk-taking in other domains, such as financial risks.
  • Uncertainty may affect people differently depending not only on their heterogeneous characteristics, but also on their pre-Covid-19 risk-taking attitudes in a given domain. For example, when it comes to online gambling, Covid-19 uncertainty may affect an avid poker player differently from someone who can barely shuffle a deck of cards. 
  • We may see cross-domain ‘spillovers’ (Dolan and Galizzi, 2015), where having to exert ourselves to reduce health risks because of Covid-19 may potentially release a risk ‘cap’ in other risk domains (for example, gambling, petty crime or online misconduct). 

The overall effect of prior attitudes and cross-domain compensations are thus be expected to make generalisable predictions even more complex.

Another practical implication of domain-specific risk-taking is that, ideally, researchers and policy-makers should make use of multiple measures of risk-taking, not just one. A growing body of evidence suggests that financial lotteries (popular tools in behavioural economics research) are often poor predictors of real-world risk-taking behaviours in other domains, such as health (Galizzi et al, 2016).

This is the main reason that economists and psychologists alike have proposed measuring alternatives (Dohmen et al, 2011). One such example is the Domain-Specific Risk-Taking (DOSPERT) Scale (Weber et al, 2002) which assesses risk-taking in five specific domains of interest: financial decisions, health/safety, recreational, ethical and social decisions. This scale is widely used and has been shown to predict real-world behaviour reliably across domains, especially in a health context (Harrison et al, 2005; Shou and Olney, 2020).

A limitation of this scale (and a limitation common to others) is that the DOSPERT is only as reliable in assessing domain-specific risk as the precise items it asks its respondents about. In other words, ‘taking a skydiving class’ (recreational risk) or ‘engaging in unprotected sex’ (health risk) are unlikely to predict the current behaviours of entire populations under lockdown. This, of course, does not mean that risks are not currently being taken in these risk domains.

Before we are able to understand, predict or integrate risk-taking in this new world, we need urgently to design, validate and replicate lockdown- and Covid-19-compatible measures of risk-taking. Such measures will function as a gateway to reproducible research and, ultimately, to answers. 

A bit of certainty

In the meantime, we should note that humans are remarkably adaptable. As we grow accustomed to the rules and measures in place to protect us, they become less attention-grabbing and, as a consequence, some of the uncertainties we face now will resolve naturally, as we adjust to a new normal.

Thus, if we are faced with an important decision right now, perhaps the most comforting thing to take away is that holding off on decisions until a form of ‘new normal’ settles in will be likely to reduce the perceived background uncertainty substantially, and bring our usual risk-taking behaviours back to normal too.

Where can I find out more?

Coronavirus and Spanish flu: economic lessons to learn from the last truly global pandemic: Chris Colvin and Eoin McLaughlin explore lessons from this historical pandemic for today’s policy-makers at The Conversation.

Back to background risk? Andreas Fagereng, Luigi Guiso and Luigi Pistaferri discuss the importance of background risk for portfolio allocation.

10 misconceptions about the 1918 flu, ‘the greatest pandemic in history’: Richard Gundermann, professor of medicine, liberal arts and philanthropy at Indiana University, writing at The Conversation.

Who are experts on this question?

Matteo Galizzi, Benno Guenther, Maddie Quinlan and Jet Sanders at the Department of Psychological and Behavioural Science of the London School of Economics:

  • Working on a systematic review of studies on risk-taking under major uncertainty; and on systematically testing fluctuations of risk-taking. 
  • Working on designing, validating and replicating a new domain-specific risk-taking scale in the time of Covid-19.

Other experts in the UK who have worked extensively on risk and uncertainty:

  • Graham Loomes (University of Warwick), Chris Starmer (University of Nottingham) and Robert Sugden (University of East Anglia), founders of the Network for Integrated Behavioural Science (NIBS).
  • Marjon Van Der Pol (University of Aberdeen), deputy director of Health Economics Research Unit (HERU), also at University of Stirling Behavioural Science Centre.
  • Magda Osman (Queen Mary University of London), head of Dynamic Learning and Decision Making Lab (DLDM).
  • Barbara Summers (University of Leeds), director of the Centre for Decision Research (CDR).
Author: Matteo Galizzi, Benno Guenther, Maddie Quinlan and Jet Sanders

Published on: 03rd Jun 2020

Last updated on: 21st Oct 2020

Do you have a question surrounding any of these topics? Or are you an economist and have an answer?

Funded by

UKRI Economic and Social Research Council
Skip to main content