Although one of the biggest risks facing a reinsurer such as Scor is mortality – the company is now as much a life reinsurer as a non-life reinsurer – there has been little research thus far into the potential knock-on effect of a major pandemic.
In a just-published paper by Michel Dacorogna from Scor and Meitner Cadena from UPMC & Crear-Essec Paris, Scor's The Art & Science of Risk latest paper, entitled "Exploring the Dependence between Mortality and Market Risks" , the writers look at six developed countries and the worst 10 years of mortality over the past 80 years.
Although the study observed a reduction in the performance of some financial variables and an increase in correlation between death rates and financial performance, the effect remained weak. In addition, the samples did not contain any significant pandemic outbreak.
The paper noted that Solvency II required a 99.5% value-at-risk estimation (1-in-200 years) while the Swiss Solvency test was at 99% (1-in-100 years). "Both cannot be reasonably computed without a good understanding of catastrophic risks like earthquakes or pandemic".
As the paper observes, if there were an increase in mortality, some of the funds currently invested on financial markets would be needed to pay those claims. However, a severe pandemic could have ripple effects on the economy that would impact the value of those investments.
"Modellers are thus faced with the difficult problem of coming up with reasonable estimates for the possible dependence between mortality risk and market risks in case of stress", the paper says.
Rather than use a prospective calculation system – the norm in most analyses to date, with the Black Death and the post World War I influenza epidemic the two favourite areas of study – the Scor study takes two sources of data, the Human Mortality Database and the Global Finance Data database.
The main pandemics in the past century began in 1918, 1957, 1968, 1981, 2002, 2006 and 2014, with two years, 1918 (Spanish 'flu) and 1981 (AIDS) both being more than six times as devastating as anything else. The former killed 30m people, while the latter killed 25m.
The study found that "the extent of the effect over five countries out of six for stock indices and for all the countries for government yield indicate that indeed mortality and market risks are correlated in the extremes".
However, the writers add that "this effect remains weak and does not allow us to conclude definitely on the size of this dependence as our samples do not contain extreme downturns in the mortality indices".