In many ways, 2020 has already been a year that has had virtually every worst-case outcome take place: a global pandemic the likes of which have not been seen since the Spanish Flu of 1918, a market crash and economic recession that were as sharp as the Great Depression of 1929, a Fed intervention that puts the 2008 bailout to shame, and topping it all off, a new cold war. But according to Deutsche Bank the "fat tail" outcomes are only getting started.
As the bank's strategists Henry Allen and Jim Reid "cheerfully" write, tail risks such as Covid-19 are thankfully rare compared with other market shocks. However, they note, such events "are a more regular occurrence than most people think, with pandemics being just one example. In this piece we consider various tail risks that could cause another global crisis, and assess the probability that they will occur."
The bank then finds there is a one-in-three chance that at least one of four major tail risks will occur within the next decade: a major influenza pandemic killing more than 2 million people; a globally catastrophic volcanic eruption; a major solar flare; or a global war.
This assumes there's a:
2% (1 in 50) annual probability of a flu pandemic killing 2.2m people
0.15% (1 in 667) annual probability of a VEI-7 volcanic eruption taking place.
1.06% (1 in 94) annual probability of a solar flare as large as the 1859 Carrington event taking place.
0.625% (1 in 160) annual probability of a global war, an assumption that is below what has occurred historically but reflects a more globally interconnected and cooperative landscape today.
If these are independent probabilities, and three of the four are natural occurrences, then there’s a 33.5% chance at least one will take place in the coming decade. And if you stretch this time frame out to two decades, then it becomes more likely-than-not (55.7%) that at least one of these events will occur. This note looks at the possible consequences of each of these events, and also considers some even more extreme tail risks that have occurred in the past that would also wreak global havoc if they occurred today.
Below we excerpt from the report's findings:
Risk #1: Pandemics
As our current crisis has shown, pandemics are one of the biggest threats we face. However, what is particularly striking about Covid-19 is that it has actually been relatively mild compared with other severe pandemics through history. The global death toll from Covid-19 so far has been over 430,000. But in fact, this is less than pandemics within living memory. For example, both the 1957-58 pandemic, known as Asian flu, and the 1968 pandemic, known as Hong Kong flu, have both been estimated to have had a worldwide death toll of over a million. That’s more than double the confirmed global death toll from the coronavirus to date. Furthermore, both took place when the global population was less than half what it is today, making the per capita impact even more severe.
In today’s environment, although improved health outcomes and scientific advances mean we are much better prepared to deal with a pandemic, we also live in a significantly more urbanised and globalised world, offering greater chances for diseases to spread.
So what are the chances of such a pandemic occurring? Before Covid-19 occurred, Madhav et al. (2017) estimated there was an annual probability of 2 per cent that an influenza pandemic will cause 2.2 million pneumonia and influenza deaths or more globally. For reference, the World Health Organization say that normally the number of deaths from seasonal influenza each year are between 290,000 and 650,000, so this is an order of magnitude above what would be expected in a normal flu season. And as mentioned, the global death toll from Covid-19 so far has been just over 430,000, albeit with strong measures in place to minimise the number of deaths.
With a 2 per cent annual probability, this means it’s more likely than not over a 40-year time frame that we’ll see such a pandemic occur. Looking at a more severe scenario, they estimate there’s a 0.2 per cent annual chance of a pandemic causing at least 22 million global deaths, which means there’s a nearly 8 per cent chance that will occur over the next 40 years. It’s also worth noting that this is just looking at the odds of an influenza pandemic, rather than pandemics from all causes.
Furthermore, we should consider the possibility of a pandemic even worse than these. Looking back through history shows some of the most severe examples, even if it seems certain that today’s norms of medicine and hygiene would have mitigated the effect if these had occurred in modern times:
A century ago, the Spanish flu outbreak led to the deaths of as many as 50 million people, more than the total number that died as a result of World War One.
The Black Death in 14th century Europe led to the death of more than a third of the European population. In England, data shows that within the space of three years from 1348 to 1351, the population declined by 46 per cent. Furthermore, the population didn’t recover back to its 1348 level again until 1622. That would be like today’s population not being reached again until 2294. It is worth noting however that the Black Death was transmitted in part by fleas and rats, so it would be harder for that to occur today.
Much further back in the 6th century, the Plague of Justinian has also been estimated to have killed as much as half of the European population at the time, though others have questioned the extent of the damage.
And in Japan, a smallpox outbreak that began in 735 has been estimated to have killed a third of the country’s population.
And much as a pandemic is primarily a public health emergency, they also have substantial secondary effects on the economy. Given that Covid-19 has led to a severe recessions and lockdowns in numerous countries, an even more serious pandemic could be even more catastrophic. Indeed, as we noted in an earlier piece looking at 800 years of economic contractions, the second largest annual economic contraction in England since 1271 was in 1349, at the height of the Black Death. Then, the economy contracted by 23.5 per cent in a single year. By contrast, DB’s forecasts for the US and Euro Area growth in 2020 stand at -7.1 per cent and -12.0 per cent respectively.
And finally, there are questions as to what would have happened if the virus hadn’t had such a disproportionate impact on the elderly, as Covid-19 has done. Had it impacted children in large numbers, one would expect that the public would be even more cautious about returning to normal life, and in turn the lockdowns and restrictions to defeat the virus could well have been more severe.
Risk #2: Volcanoes
Another of the biggest risks to global stability comes from volcanic eruptions. Consider that the relatively remote Eyjafjallajökull volcano in Iceland shut down nearly all of European airspace in 2010, leading to extensive economic disruption. And further back in 1991, the eruption of Mount Pinatubo in the Philippines was so large it actually led to a reduction in global temperatures over the subsequent two years.
So an important question is what would happen if a much larger, more catastrophic eruption occurred, and how likely is this to happen? To frame this, we use the Volcanic Explosivity Index, which measures the explosiveness of volcanoes. A VEI- 5 event means there’s been at least 1 km3 of ejecta volume, and we’ve seen one of these in the 21st century so far. A VEI-6 event is where at least 10 km3 erupts, and the last of these was the Pinatubo eruption in 1991.
A VEI-7 event, with over 100 km3 of volcanic material would cause major disruption on a global scale. The primary effects would be life-threatening consequences for any people located nearby, with risks of a very high death toll if an eruption occurred near any major cities. As the fallout spread across the world’s atmosphere, heavy restrictions on air travel would be implemented which would severely affect the tourism and hospitality sectors, similar to the experience seen during Covid-19. As global temperatures and weather patterns changed, crops and food production would be affected leading to increased prices and humanitarian disasters, particularly in emerging and frontier countries.
To estimate the probability of such an event, Newhall et al. (2018) write that frequency of VEI-7 events is between 1 and 2 per thousand years, though probably closer to 2. If we assume there’s a 0.15% annual probability that such an event will occur (or around once every 667 years), then the chances that something will happen over the next 40 years are 5.8%. So an unlikely occurrence, but far from an impossible one.
The last time a VEI-7 volcanic eruption occurred was in 1815, with the eruption of Mount Tambora that’s located on Sumbawa, Indonesia. The eruption itself could be heard over a thousand miles away, and killed tens of thousands of people located nearby. However, the consequences spread from Southeast Asia to the entire world.
The primary effect was extensive climactic disruption. The volcanic ash and sulphur that was spewed out led to what is known as a volcanic winter, where global temperatures were lower as a result of the material from the volcano obscuring the sun. In fact, the following year in 1816 was known as the “Year Without a Summer” because of this reduction in temperatures.
The changes in the climate led to crop failures throughout the world, creating widespread hardship and famine, which in Europe coincided with the slump following the Napoleonic Wars. Indeed, the events of these years have been linked to a number of food riots across the continent, as hungry and desperate populations struggled to afford to eat.
Looking further back, another VEI-7 eruption was that of Mount Samalas in 1257, also in present-day Indonesia. Though the evidence is much less extensive given the time periods involved, it has been estimated that the eruption similarly caused extensive global cooling. Vidal et al. (2016) write that the eruption “released enough sulphur and halogen gases into the stratosphere to produce the reported global cooling during the second half of the 13th century, as well as potential substantial ozone destruction.” Meanwhile Oppenheimer (2003) points out a summer cooling pattern in both hemispheres in 1258 and 1259, the years following the eruption.
So twice in the last millennium we’ve seen a volcanic eruption large enough to cause extensive global cooling, as well as vast economic and political disruption in multiple continents. Given the extent of globalised supply chains and modern aviation capacity, the risks from such an eruption today could be even greater.
Risk #3: Solar Events
This next risk comes from millions of miles away – the sun itself. A severe solar storm can cause substantial and long-standing damage here on Earth. Indeed on a much smaller scale, we’ve seen how this can occur when a geomagnetic storm in 1989 caused the loss of power in Quebec for nine hours and the Montreal Metro was shut down. Another solar flare later that year led to a trading halt on the Toronto Stock Exchange.
These events, though, were relatively contained. Far more damaging would be something akin to the Carrington Event of 1859, so named after the astronomer who recorded it. This is the biggest geomagnetic storm on record, and caused sparks to leap from telegraph lines that were in usage at the time causing shocks to the operators. The Northern Lights could even be seen as far south as Cuba.
A similar event occurring today would be far more disruptive given that modern technology is based on electronics. There could be major power outages as electrical power grids are disrupted, which in turn would have knock-on effects throughout the economy as critical infrastructure is unable to be run properly. Lives could be lost if it impacted hospitals and medical care. Communications would be disrupted, many payment systems would be dysfunctional, and GPS satellites would face extensive interference, to the detriment of all the individuals and industries that rely on accurate location services, not least aircraft. Infrastructure that runs the world’s financial markets could be heavily damaged.
In terms of the chances of such an event occurring, Riley (2012) estimated that the probability of another event the size of the Carrington event of 1859 occurring is around 12 per cent over the coming decade. 7 So if this probability remains constant from decade to decade, then there’s a 40% chance that this will take place in the next 40 years. And stretching the timeframe to 55 years would push that probability just over 50%.
Looking even further back, worse scenarios have occurred. There’s evidence that in 774-5 a major solar event took place from spikes in radiocarbon from tree rings. Mekhaldi et al. (2015) imply that the 774/5 event “was at least five times stronger than any instrumentally recorded solar event.” And Sukhodolov et al. (2017) conclude that this “was able to decrease the stratospheric ozone for more than one year and thus to modulate the surface weather.” More recently in 2012, the Earth experienced a near-miss from a major solar storm, which would have been the largest in over a century had it hit.
Since we haven’t had a Carrington-level event in the last 150 years, it’s difficult to tell with precise certainty what it would mean for today’s technology. But it’s clear that the effects of solar storms on Earth could leave a long legacy of damage akin to other natural disasters seen more regularly through history, especially in a world so reliant on electricity and other modern technologies.
Risk #4: Outbreak of war or a nuclear accident
We have been fortunate to live in an unusually peaceful period by comparison with most of history, with no major global war having taken place for 75 years. Unfortunately however, this is historically atypical, as an examination of recent centuries will demonstrate.
Only in the last century, World War II led to the death of around 3 per cent of the global population at the time, while World War I led to the death of around 1 per cent. Furthermore, a number of more localised conflicts, such as the Russian, Chinese and Spanish civil wars, led to substantial numbers of deaths in the countries where they took place. And looking further back into the 17th, 18th and 19th centuries shows a litany of conflicts both between and within the major global powers.
Today there are fortunately a number of reasons why wars are less likely than previously. There are many more forums for communication and cooperation between nations, which are also far more economically interdependent on one another. In a nuclear world, the gains from war are much more limited since using such weapons risks mutually assured destruction, something that helped to prevent an escalation during the Cold War. And the democratic structures that have replaced absolute monarchies throughout much of the world mean that we rarely see wars today taking place simply due to an ambitious monarch’s desire for territory.
Nevertheless, though this offers some consolation, many conflicts of history have not begun with both sides actively seeking war, and there have also been a number of near misses. This suggests we should not be so complacent:
World War One: Although the assassination of the Archduke Franz Ferdinand was the tipping point that set the world on the path to war, it was initially a localised incident between Serbia and Austria. In fact, efforts over the month after the assassination sought to avoid a conflict.
Cuban Missile Crisis, 1962: This was a similar crunch point that didn’t lead to the outbreak of nuclear war, but has widely been considered an incident that came close to one. Indeed, US President John F. Kennedy subsequently said that he thought that the odds of a nuclear war at the time were at least one-in-three.
1983 Soviet false alarm: In 1983, there was a Soviet false alarm of incoming US missiles. Fortunately, these were not escalated by the officer on duty, but if they had been, we can only speculate what might have resulted. This is particularly so given the speed of any nuclear attack would give literally minutes to decide on a response.
Unlike the previous risks, all of which were natural phenomena, it is more difficult to place a probability on the outbreak of war, since they are subject to human calculations. However, if we made the conservative assumption that the annual probability of a global war is 0.625% (i.e. a 1 in 160 chance each year), then that means there’s a 22 per cent chance that one will take place within the next 40 years. As mentioned at the beginning, this probability is some way below that seen through history, but accounts for the fact that we live in a far more interconnected and cooperative global environment than in centuries past.
Risk #5: Earthquakes
The risk of an earthquake is a more localised one, hence we haven’t included it in our initial probabilities that specifically look at global crises. Nevertheless, the potential damage and consequences that have arisen from earthquakes through history can be extensive.
For example, less than a decade ago the 2011 Japanese earthquake was the fourth most powerful since records began, and led to over 15,000 deaths. The Nikkei 225 index plunged by more than 15 per cent in the immediate aftermath, and the subsequent nuclear accident at Fukushima saw other countries move against nuclear power. For example in Germany, the government said that they would close their nuclear power plants by 2022.
One region that is vulnerable in particular is the west coast of the United States. The US Geological Survey estimate that within the next 30 years there is a 31 per cent probability that an earthquake measuring at least magnitude 7.5 will occur in the Los Angeles region, and a 20 per cent chance of one occurring in the San Francisco region. Raising the severity of the earthquake to magnitude 8.0 or higher gives odds of 7 per cent for the Los Angeles Region over the next 30 years, and 4 per cent for the San Francisco region.
These have been seen in the past. The 1906 San Francisco earthquake (of magnitude 7.9) caused extensive damage to the city at the time and led to over 3,000 deaths. Fortunately building designs are much stronger today, but a large earthquake is likely to happen at some point in the region, which will have major implications for the people and companies operating there. In particular, a severe earthquake would disrupt many technology companies with consequential impacts for many of those listed on the Nasdaq. Although the covid-19 pandemic has shown that many of these companies can operate with remote staff, one reason why companies in the area have performed so well is that they benefit from the proximity of the people for idea sharing, investment, and job opportunities. If an earthquake eliminates this proximity advantage, US technology companies could experience a medium-term setback.
The probability of a crisis
Taken together, there is a one-in-three chance that over the next decade, one of four major global crises will occur: a major influenza pandemic killing more than 2 million people; a globally catastrophic volcanic eruption; a major solar flare; or a global war.
Running through the assumptions behind this that we discussed in the piece:
Madhav et al. (2017) find using exceedance probability analysis that there’s an annual probability of 2 per cent that an influenza pandemic will cause at least 2.2 million global deaths.
Newhall et al. (2018) write that frequency of VEI-7 events is between 1 and 2 per thousand years, though probably closer to 2. We therefore assign a 0.15% annual probability to such an event occurring.
Riley (2012) estimated that the probability of another solar storm the size of the Carrington event of 1859 occurring is around 12 per cent over the coming decade.
And we make the historically conservative assumption that there’s a 0.625 per cent annual probability of a global war, i.e. a 1 in 160 chance it will happen in a given year.
Assuming that these are independent events, then the probability that at least one of these will occur in the next decade is just over one-in-three (33.5%). And therefore stretching the time frame to a 20-year period means that it is more likely than not (55.7%) that at least one of these events will occur.
To put these odds into context, our one-in-three chance of a global crisis in the coming decade is higher than the odds PredictIt puts on the Republicans winning back the US House of Representatives in November’s election. Or in another sphere, it’s more than the odds on Betfair’s Exchange that the current favourites Bayern Munich have of winning this year’s Champions League.
The Interaction of Shocks
While we have examined these tail risk events in isolation, large shocks normally have second-round effects after the initial impact has passed. They also interact with other shocks and processes that occur at the same time. For example, the Covid-19 pandemic has been a pretext for an escalation in tensions between the United States and China. Meanwhile unemployment has risen to multi-decade highs in a number of countries, particularly for the young. Furthermore, the delay in treatments to non-covid illnesses is likely to lead to a wave of more serious conditions. Though we are yet to fully understand the long-term effects that the coronavirus will have on the economy and society, it is clear these will stretch well beyond the immediate health effects of the pandemic.
Indeed, just look at how political turmoil and revolution has regularly followed economic disruption. For example, the 1929 crash led to austerity policies that amplified the depression and had a devastating impact on the poor. That helped fuel several aggressive political movements. Further back, the French Revolution of 1789 came after a succession of poor harvests that resulted in sharp rises in food prices. Later on, the 1848 Revolutions in Europe similarly came at the end of the so-called ‘Hungry Forties’ and severe famine in much of Europe.
In fact, there are entire periods of history that have been collectively labelled crises. In the Ancient World, the so-called “Crisis of the 3rd Century” saw the Roman Empire experience extensive problems – including the Cyprian Plague – as the empire broke into three. Later on, in the 14th-15th centuries, the combination of famine, plague and conflict in Europe has seen the events of the period contextualised into a broader “Crisis of the Late Middle Ages”. And the chaos associated with the 17th century – which saw a number of major conflicts take place - has led some historians to describe it as “The General Crisis”. It is to be hoped that the 21st century doesn’t also gain its own ‘crisis’ epithet.
The biggest tail risks of all…
So far in this piece, we’ve stuck to events that could plausibly occur in the coming decade. But if you’re looking for some even bigger tail risks, albeit much less likely, we briefly consider a couple of possibilities we’re unlikely to see in our lifetimes.
One possible extreme tail risk is a VEI-8 volcanic eruption, which would be much more severe than the VEI-7 cases discussed earlier. One example of this is the eruption of Mount Toba around 75,000 years ago. Rampino and Self (1993) write that this could have led to a volcanic winter with temperature declines in the Northern Hemisphere of around 3 to 5 degrees Celsius. Though Oppenheimer (2002) puts the figure closer to 1 degree Celsius, he also writes that the eruption “might have played a role in triggering a millennium of cool climate”. 12 Depending on the precise regional impacts, this could have vast implications from questions of food security to mass migration.
Another possibility is an asteroid strike, the most famous of which occurred around 66 million years ago and led to the mass extinction of the dinosaurs and a host of other life on earth. While this may seem implausible, smaller instances have happened in recorded history. In 2013, the Chelyabinsk meteor exploded over Russia with the power of an atomic bomb. Thankfully, the location of the explosion meant there was limited damage. In 1908 in Siberia, an explosion that is thought to have been caused by an asteroid that exploded in the sky flattened around 2,000 km2 of forest, known as the Tunguska event. Earlier still, in 1490, the Ch’ing-yang event occurred in China where contemporaries recorded stones falling from the sky like rain, and has been estimated to have led to thousands of deaths. Though unlikely, this is again an event that could lead to a severe crisis if it were to occur in a heavily populated area.
We’ve been fortunate to live through a comparatively stable period of history. But the coronavirus has served as a reminder that our economies and societies can change rapidly overnight. Tail risks that can cause these changes are more common than many realize and it is more likely than not that the coming decades will see another with consequences at least as severe as those caused by Covid-19.