The potential for devastating pandemics is still there, with major accumulation risks for insurers and reinsurers increasingly relevant for non-life lines of business. By David Benyon
A major pandemic could create unmanageable risks of accumulation for insurers and reinsurers, one century since the biggest pandemic of the modern era, the Spanish influenza outbreak at the end of the First World War, devastated 20th century humanity.
Today’s world has globalised by international air travel and worldwide supply chains; it is medically advanced; and society is well-informed by constant and social media updates. These conditions provide new vulnerabilities as well as strengths, that would make a pandemic comparable to Spanish flu play out very differently.
As pandemics go, Spanish flu was the big one. The influenza outbreak killed more people than the Great War did in four years of mud, blood and trenches.
From Pacific Islands to the Arctic, it killed more people in 24 weeks than AIDS claimed in 24 years. It infected perhaps 500 million people worldwide in 1918-1920, killing perhaps 100 million people, up to 6% of the world’s population at the time.
Its name, incidentally, came from censorship in affected countries suppressing news of the outbreak, while Spain was one of the few neutral countries without wartime reporting restrictions. Even then, media played an important role in influencing pandemics.
The H1N1 virus was deadly because unlike regular seasonal flu outbreaks it targeted those whose immune systems were strongest, killing off millions of young adults, many of whom were itinerant refugees crossing regions and borders, and millions of soldiers returning home overseas from the global conflict. In the US in 1918-1919, some 99% of pandemic flu deaths occurred in people under 65, and nearly half among those between 20 and 40 years old.
The most recent pandemic was an H1N1 swine flu outbreak, the only pandemic of the 21st century thus far. The virus was first identified in Mexico in April 2009 and spread rapidly worldwide because it was a new type of flu, to which few young people were immune, rather than seasonal flu which tends to kill the weakest – the very young and the old. It became known as swine flu because of its similarity to flu viruses affecting pigs. The outbreak was not as deadly as feared, partly because many older people were already immune to it.
Before Swine flu, and after Spanish flu, pandemics struck several times killing further millions of people during the 20th century. Asian societies sustained the heaviest casualties. In 1956-1958, an H2N2 Asian flu pandemic spread from China to reach Hong Kong and Singapore, causing 2-3 million deaths. It had also reached the US by June 1957, where it killed nearly 70,000 people.
Hong Kong was hit again by H3N2 strain by an outbreak of the influenza A virus in 1968, along soon thereafter with China, Vietnam, Philippines and Singapore. Within a few months the virus had reached India, Australia and Europe, and later Japan, Africa and South America. British newspaper The Times helped raise the alarm, while governments were little better prepared than they had been a decade earlier. The category-2 severity pandemic mostly affected older people, claiming about one million deaths globally.
Other recent outbreaks that have fallen short of pandemics were the 2002-2004 SARS epidemic which spread from China, and the H5N1 bird flu outbreak, which spread from China in 1997. Luckily it was not spread by human to human transmission, limiting its impact.
In West Africa and Latin America, outbreaks of the Ebola and Zika viruses in recent years have also hit the headlines, causing deaths and disruption. Ebola was unlikely to become a pandemic because its spread is via bodily fluids rather than in the air, while the Zika fever was spread by mosquitos.
Understanding the fallout
While a huge pandemic on the scale of Spanish flu has not occurred within current generations’ lifetimes, several smaller pandemics as well as more epidemics have happened in recent years.
Willis Re’s head of life and health reinsurance Greg Solomon warns not just about pandemics but “mini-pandemics” that could become more frequent, particularly as viruses become resistant to drugs. “We don’t know when the next one-in-200-year meltdown will happen. We already had a one-in-100-year event with the financial collapse of 2008,” says Solomon.
In risk and capital terms, regulators have made allowance for this. Life catastrophe risk is explicitly to be modelled under Solvency II’s standard formula for life insurance.
Irfan Akhtar, head of Aon ReMetrica for life, health and pensions, says: “On average we’ve seen a pandemic once every 25 years. Solvency II uses three in its risk assumptions – one every 33 years – but it could be four or five. The severity could vary greatly, but it is seen as a rare event.”
There is a temptation to write off a major pandemic as a once-in-a-century or even more infrequent occurrence. Focusing on an apocalyptic pandemic scenario can – understandably – give way to fatalism, because firms might think the scale of the disaster is so big but so low frequency that it is not worthy proper preparation.
“They think if it bankrupts them, it will also bankrupt their competition, and they’ll be dead anyway, so why prepare? But what they should prepare for is a one-in-fifty-year pandemic. For that you can buy reinsurance cover or capital markets protection, so the impact would not be as bad as it might be.”
From a mortality risk perspective, pandemic risk is huge but relatively well considered. “Does the insurance industry consider it? I would say certainly,” says Gunther Kraut, head of epidemic risk solutions at Munich Re. “Pandemic is recognised as one of the peak risk scenarios, much like earthquake events. Solvency II calibration provides one way for internal models to include pandemics, and you have external vendor risk models,” Kraut adds.
However, on the non-life side its consequences may be less well understood. According to Kraut pandemic’s broader consequences remain difficult to model. For a modern-day Spanish flu-strength severity pandemic, a different picture would emerge for insurers, with accumulations and correlations across all lines of business.
The life side to pandemics is better understood. “Ebola in 2014 was a big humanitarian catastrophe, but it was also an economic catastrophe,” Kraut says. “Many people lost their jobs, construction, airports, ports and supply chains were interrupted. Mining companies were badly affected, and some smaller ones went into bankruptcy.”
Accumulation risk could hit companies, conglomerates, insurers and reinsurers in ways they have not yet been considered. For non-life insurance companies, absenteeism among staff would not be only problem, Willis Re’s Solomon explains.
“On the balance sheet, half the assets and liabilities would be hit at the same time. Mortgage defaults could soar, and banking would be badly hit. This is a systemic risk, because it’s all correlated. A one-in-50-year pandemic is a terrifying scenario,” says Solomon.
There has been increased interest among companies for business interruption and loss of earnings covers for such risks. Tourism – through cancelled trips and flights – is an obvious line of business, particularly for parametric trigger products.
“There are tourism industry discussions around protection if revenue drops,” says Akhtar. “Similar to that, for the US private health sector, planned operations are profitable. They lose revenue if a pandemic happens and planned operations are cancelled because of sudden demand to deal with the pandemic, so they can be compensated by insurance.”
One of the challenges is lack of data to model. “We don’t have many pandemics to analyse, which means a lack of tangible numbers,” says Akhtar. “Things tend to come about that we never predict. For example, one 30,000-year-old virus came back to life when the Siberian permafrost started to melt. Luckily it didn’t affect humans, but it might have done.”
Munich Re is among those trying to model pandemics to design re/insurance products.
“The first thing is to understand and model the risk,” Kraut says. “Then you have to manage accumulation risk. We collect historical data from the World Health Organization (WHO) other publicly available sources and build up our actuarial risk analysis.”
The unit started as an innovation project, and now offers life and non-life covers.
“Since the West African Ebola outbreak, we offer a parametric-based pandemic emergency facility to the public sector,” Kraut says. “In that case the World Bank acts as the counterparty for the fund, mostly provided by G-7 countries’ aid and development budgets, to perform and execute this transaction, bought on insurance markets and as a cat bond.”
He adds: “The feeling was the international community could do better to contain an outbreak. Since then three outbreaks have qualified to be closely monitored: two Ebola outbreaks in Congo, and a Lassa fever outbreak in Nigeria.”
Globalisation and the ubiquity of international air travel have transformed the way a pandemic of comparable lethality to Spanish flu – or even a lesser one – would spread in today’s world. Geography is less of a barrier from this risk than it was in 1918. With four billion air travel journeys per year, pandemics now spread faster and differently. A century ago, when a journey from the UK to Australia might take longer than a month, nowadays it takes a day to get from London to Sydney.
For example, Dirk Brockmann, a professor at Berlin’s Humboldt University, has posted on Youtube about “the hidden patterns in complex global disease dynamic”. a video clip depicts a computer simulation of a global pandemic. The spreading takes place on the worldwide air transportation network of more than 4,000 airports and 25,000 direct flight links. The pandemic leapfrogs continents in hours.
“We started to see this change in 2009 with swine flu, which went from Mexico to China and hit at least twenty countries in three weeks.” says Akhtar. “In 1918 the spread or movement of disease was intuitive like that of concentric rings spreading uniformly in all directions from a single starting point. Today, passenger air travel means the effective distance between neighbouring small towns becomes relatively further away than between cities, such as London and New York. There can be more connectivity between big cities than between towns and villages in the same country.”
Akhtar notes that travel is set to massively increase. Not only that, but the aviation centre of gravity is shifting eastwards, towards populous African and Asian countries. “It is set to double in the next 20 years to 8.2 billion passengers annually by 2037, and to double again to 16 billion passengers by 2050, according to an IATA estimate,” he says.
Human behaviour is crucial to understanding pandemics – and it is hard to model. “Hysteria is hard to predict,” says Akhtar. “People rarely fake death, mortality happens if it happens, but it’s inherently harder to predict customer or employee behaviour.”
For example, relatively small outbreaks of lethal diseases can lead to a disproportionate reaction in human behaviour, including cancelling flights and eschewing journeys, and staying indoors, causing considerable economic losses. You just can’t count on humans to behave rationally.
“The perception of outbreaks is food for thought,” Kraut says. “It’s something we often consider because certain small outbreaks catch a lot of attentionwhile important endemic diseases like the flu seasons are always there but they are perceived differently.”
Pandemics can be defined as epidemics that have spread across countries or continents, but even the WHO has had a hard time deciding when to classify an outbreak as a pandemic, such as in 2009. Doing so means enacting countries’ pandemic preparedness plans (PPPs).
Acting too late means being blamed for the death toll but sounding the alarm too early risks crying wolf and criticism for triggering costly PPP schemes, including quarantines, border controls, and extra resources to hospitals.
Even the data experts at Google have fallen foul of this. The web giant pioneered an influenza-based prediction in 2008, based on search engine queries. If searches crossed a threshold, it could predict a pandemic. In 2009 it was two weeks ahead of the WHO. However, seasonal flu outbreaks beat its algorithms in 2012-2013, creating a false prediction.
While health systems would react much quicker in 2018 than they did in 1918, a 24-hour press and social media mean that populations are also far better informed. This might reduce the life fallout, but it could add to non-life insurance losses, such as those related to business interruption.
“You can communicate instantly through online news and social media, which would help contain a disease,” says Akhtar. “For life and mortality, communications channels are of benefit, but things linked to people’s behaviour are also likely to be affected. Social media pandemonium would make people act according to their worst fears, so the economic impact would be much greater.”
This happened with SARS in 2003, for example, when a small number of fatalities caught popular attention and led to huge economic costs. The shift towards remote working could be a major benefit in a future pandemic scenario, Akhtar thinks. The same thing happened with the 2015 Zika virus outbreak in Latin America. “When something’s out of the news, out of sight means out of mind. In the case of Zika, a lot of people cancelled their honeymoon trips to the Caribbean, and many athletes avoided the Rio Olympics for the same reason” Akhtar adds.
Akhtar sums up: “There can be an unfortunate trade-off when trying to contain a pandemic by alerting the public: having better protection in place, by which the life market is less affected, but the non-life side can be worse affected by public hysteria. During a pandemic, accumulation of risk is always a major consideration for insurers and reinsurers.”
By David Benyon