Were insurers and reinsurers caught out by failing to predict the risk of the pandemic?

The spread of infectious diseases figured prominently in the risk maps identified, studied and modeled by insurers and reinsurers long before the Covid-19 pandemic. SCOR developed its own pandemic risk quantification model as early as 2007. In recent years, prudential regulations have also helped (re)insurers to develop their knowledge and modeling of this risk. In line with the prudential regulations in force in Europe since 2016, (re)insurers must calculate the cost of a bicentennial pandemic – i.e. with a probabilistic return period of 200 years – and show that they can absorb that burden. This is one of the solvency tests. In 2013, SCOR issued a catastrophe bond to cover itself against a significant increase in mortality in the United States. And we organized an international conference on the risk of a pandemic with the Institut Pasteur in June 2018! In short, monitoring and managing pandemic risk was already an integral part of our risk management when the Covid-19 epidemic broke out. 

(Re)insurers were therefore not “taken by surprise” by the shock of the pandemic, which was well within the "area" of the scenarios they study and model, with a risk that was quantified in terms of both probability and severity. Nevertheless, they did underestimate the impact of the decisions taken by governments – such as lockdowns, curfews and administrative closures – to curb the spread of the virus. These measures, which have ultimately limited or halted economic activity and caused considerable operating losses for businesses, have very significantly impacted the exposure of insurers and reinsurers to the Covid-19 crisis. Consequently, to date, nearly 90% of the cost of the pandemic concerns Property & Casualty reinsurance, while only 10% relates to Life & Health (re)insurance. This situation is virtually the opposite of what had been anticipated and modeled up to now! The decisions taken by governments have, de facto, profoundly transformed the nature of the risk for (re)insurers. 

Wasn’t this government intervention predictable? 

How do you model political decisions? I would like to stress that if there was one party who was taken completely by surprise, it was the public authorities: nothing was ready! So how are we supposed to have been able to anticipate what they were going to do? Unlike natural disasters, these decisions do not obey "laws of nature". They are stochastic, a bit random, changing, with a lot of trial and error. And the fact that the measures taken in response to the pandemic have been very different from one country to the next – when everyone is facing very similar situations – bears witness to this! Let’s remember here that certain countries have had no lockdown. Even in the Europe of 27, there have been 27 different ways to manage the crisis! And within a given country, a political changeover can completely alter the measures implemented. How, for example, could we model the outcome of the presidential elections in the United States, and quantify, ex-ante, its impact on the implementation of the vaccination campaign across the Atlantic? It’s absolutely impossible to model administrative and political decisions, in the broad sense of the term, because those decisions involve a drastic and intractable element of uncertainty.

As a (re)insurer, what lessons have you learned from this crisis? 

As a reinsurer, the most significant lesson to come from this crisis is that Covid-19 has been much more of a Property & Casualty (re)insurance shock than a Life & Health one in terms of claims. 

Another “surprise” is the fact that (re)insurers have been vilified by many commentators and policymakers for not offering cover against the economic consequences of the pandemic, particularly operating losses due to government-ordered shutdowns and lockdowns. If the insurance market does not offer such protection, it’s obviously for fundamental reasons, which are linked to the very nature of the risk. This risk is intrinsically uninsurable, for three main reasons. 

First of all, it is a “serial” risk. Virtually all economic agents are affected at the same time, which creates a clear problem of risk pooling, whereas this is the cardinal principle of insurance. A corollary of this serial nature is that the size of the risk is colossal. How can you "insure" a significant percentage of a country's GDP?

Second, this risk has a very strong endogenous component, insofar as it is very dependent on political decisions. Its magnitude is determined by a myriad of variables directly controlled by the public authorities: whether or not to implement a lockdown; if so, for how long; which businesses can continue to operate, fully or partially, and which must be shut down entirely; the nature and extent of crisis buffers – such as furloughs and short-time working – put in place to avoid tearing at the economic fabric of a country, and so on. The problem is that these political decisions are intrinsically neither anticipatable nor modelable, which makes it impossible to assess the risk involved and to calculate an insurance premium.

Finally, the last, more technical, reason for the non-insurability of this risk is linked to potential issues of adverse selection and moral hazard. Adverse selection is the effect whereby only the most exposed economic agents take out insurance, which reduces the risk pooling effect. In this case, only companies in the most affected sectors - hotels, restaurants, tourism, etc. - would buy protection (unless it were compulsory, which would raise questions of acceptability). Companies in sectors that have been spared – such as e-commerce – would have no interest in seeking insurance. Moral hazard is the effect whereby the risk is increased by a behavioral change in economic agents once insurance plays the role of safety net. In this case, coverage of the cost of lockdown by third parties (insurers) would create an obvious moral hazard issue for governments, which would no longer assume the economic cost of the decisions they take, even though they are deciding the ways and means to respond to the crisis and are ultimately responsible for managing it! Such a situation could actually encourage those in power to implement more economically costly measures, the unbearable burden of which would fall on insurers and reinsurers.

The French Finance Ministry and insurers have nonetheless devised a system that would allow risks to be pooled, but which has failed to materialize in the face of mandatory contributions 

To date, no country in the world has managed to develop a system that covers business interruption losses in the event of a pandemic. This is not surprising. It’s not about “bad faith” on the part of insurers, it’s a question of technical and economic impossibility. The risk of the economic consequences of a pandemic is serial – in both space and in time -, endogenous, unmodelable, and subject to moral hazard and adverse selection. All of which means that it is not insurable. I wrote this at the beginning of this crisis – more than year ago – and I have yet to find anyone who can demonstrate that it’s not true. 

Does cyber risk share any common traits with pandemic risk? 

There are indeed many similarities between these two risks. It's no coincidence that the term "virus" is also used in IT! Cyber risk is also serial, notably due to the rise of connected devices. Like pandemic risk, cyber risk is virtually impossible to localize. And it is also endogenous: it can be changed drastically by the prevention measures implemented (or not) by policyholders. Finally, although it is not actually unmodelable, cyber risk is difficult to quantify and to model, not just because it is endogenous but also, more fundamentally, because of the lack of data involved and its ever-changing nature, which limits our ability to assess it prospectively based solely on the observation of past losses. 

How do you respond to insurers who criticize reinsurers for being more cautious in the face of growing risks? 

We should remember something obvious here: (re)insurers create value by (re)insuring, so accusing (re)insurers of not wanting to insure risks out of "cautiousness" or "bad faith" is as absurd as accusing a baker of not wanting to sell more bread! If the (re)insurance market does not offer protection for a given risk, it’s simply because the insurability conditions for that risk have not been met.

The will - and vocation - of (re)insurers is to push back the frontiers of insurability. However, it must be stressed that expanding the sphere of insurable risks is not down to (re)insurers alone. Policyholders – and risk managers – also have an essential role to play in collecting and sharing information, exchanging expertise, quantifying damage, and actively implementing preventive and control measures, in order to reduce the probability of the risk occurring and to control its severity if it does occur. Otherwise, (re)insurance supply can only structurally be lower than (re)insurance demand. This imbalance results in a sub-optimal situation, where (re)insurance is very expensive - with restrictive coverage conditions - if not totally unavailable. Meeting the conditions of insurability in the face of modern risks therefore requires a kind of joint risk management between risk carriers - the (re)insurers - and those facing the risks. This sharing of responsibilities is an absolute prerequisite for positively shifting the frontiers of insurability and thus reducing the protection gap, which is still substantial on a global scale.

Will the 21st century be the century of catastrophes? 

De facto, the risk universe is both expanding and changing. It is constantly growing and transforming, with technological innovation and scientific progress on the one hand, and globalization and the development of human activities on the other, which leads to the emergence of increasingly complex interdependencies. The risk universe therefore needs to be understood and analyzed in a holistic way. Risks are increasingly serial and global. Many risks are no longer restricted by time and space, as was traditionally the case. 

But we must be careful not to sink into disaster mongering! We must remain confident in the capacity of human intelligence to manage risks and meet the related challenges. I am fundamentally a positivist; I would even say a Comtist. We are living in an extraordinary and very fertile period in terms of science and technology: quantum computing, artificial intelligence, big data, machine learning, robots, blockchain, nanotechnologies, genetic research, etc. Even if they multiply the risks that our societies must face, all these developments are intrinsically very positive. I must stress that the principle of prevention is more important than the principle of precaution. If the latter means that we should refrain from developing technologies whenever there is uncertainty over their consequences, then the precautionary principle is itself dangerous, as it could slow down scientific progress and innovation!

The right attitude is to actively manage risks at all levels: to promote a culture of risk; to generalize risk management at both an individual and collective level; to invest in understanding risks; to develop preventive measures that reduce the likelihood of risks occurring as well as their severity; and to promote risk sharing, including through public-private partnerships. The 21st century will be the century of risk management - or not.  

What is your scenario for 2021?

The Covid-19 crisis is still ongoing, and it’s impossible to predict with any certainty how it will develop in the future. SCOR has developed a compartmental epidemiological model that allows us to model the development of Covid-19-related mortality according to several assumptions - such as the duration of lockdowns, how strictly social distancing is observed, the speed with which vaccination campaigns are implemented, and the effectiveness of the vaccines - and to measure the impact of these different parameters on the severity and duration of the pandemic in different regions of the world. Based on this model, which we have been constantly improving for more than a year now, and using a plausible range for the different parameters, we expect a very significant reduction in the number of pandemic-related deaths by the end of the third quarter of 2021. However, the veracity of this projection depends very much on the implementation of vaccination campaigns and their acceptance by the population - which are still subject to many unknowns - as well as the possible emergence of new variants.

Looking more at the economy, are worried about the return of inflation? 

No one is in a position to predict whether an acceleration in inflation, and interest rates, will occur. What is clear is that the likelihood of such a scenario has increased significantly in recent months, particularly in the United States. We are not afraid of this scenario at SCOR, for a simple reason: the Group is fully prepared! 

On the assets side, we are pursuing a prudent investment strategy, with a limited appetite for market risks. Our bond portfolio, which constitutes the bulk of the Group's investments, is of very high quality, with an average credit rating of A+, and benefits from a short duration. As a result, our investment portfolio is very liquid, with EUR 8.4 billion of cash flow expected over the next 24 months, which could be reinvested at more advantageous market conditions if inflation, and hence rates, were to rise.

On the liabilities side, we are only very slightly exposed to inflation risk. Our Life reinsurance commitments, which focus on the coverage of so-called “biometric” risks - mortality, longevity, long-term care, health, etc. - are not affected. Our Property & Casualty commitments would certainly be impacted by an acceleration of inflation, but in a contained way insofar as our risk portfolio is essentially made up of short-tail rather than long-tail lines.

Overall, the rise in interest rates would be good news for the SCOR group, which would see an increase in both its solvency and its profitability.
I would add that, in a scenario of a sudden rise in inflation and interest rates, SCOR's liabilities would be exposed to almost no liquidity risk. As far as our Property & Casualty (re)insurance business is concerned, there are two reasons for this. On the one hand, this business is structurally cash flow positive, as it benefits from a reverse production cycle: premiums are collected before payment of any claims. On the other hand, the corresponding liability on our balance sheet is highly "viscous": it can only be "withdrawn" if our (re)insured clients suffer a loss entitling them to compensation. The Property & Casualty (re)insurance business is therefore not associated with the risk of a "bank run" - i.e. a sudden mass withdrawal of liabilities - that could generate a liquidity crisis. Similarly, in our Life reinsurance business, we are not exposed to liquidity risk because of our exclusive focus on biometric risks. 

The government has just decided to extend the restrictive measures for the month of April. What do you think of the public authorities' response, which has prioritized life over the economy? 

The decision by the authorities to focus on health rather the economy is obvious, as evidenced by the now famous expression “quoi qu’il en coûte” (whatever it takes). Other countries have made a more balanced and judicious choice. 
Once again, what strikes me is on the one hand the unpreparedness of the public authorities to deal with the health crisis, and on the other their blatant lack of efficiency. The pandemic took all the institutions involved in France by surprise: hospitals were unprepared, screening capacities were almost non-existent for a long time, masks remained unavailable for a long time, lockdown procedures had not been planned, and so on. And today, we are facing a shortage of vaccines! 

Public risk management has obviously been deficient. We had already seen this before the health crisis, as demonstrated by the chronic public debt situation the country has been in for more than 40 years. For too long, the government has limited itself to a mainly curative, reactive role. And it always seems to act in haste. It must now prioritize prevention and protection. It’s perfectly legitimate for the government to organize prevention, to ensure the safety of its citizens. This is part of its sovereign responsibility. In the future, the government will need to make greater use of technology to be more efficient.

Has the world changed with this crisis ? 

The violence and universal nature of this shock, both in terms of its globality its multi-dimensional consequences – health, social, economic, financial and geopolitical – will obviously have a huge impact in many areas. I wrote a year ago that this shock could significantly and lastingly shift the “axis” on which the world has turned until now. This belief has not changed and has even grown stronger: the course of History will be substantially altered by this crisis, which is challenging many of our beliefs and calls for original solutions in many areas. We are going to have to reinvent the world!

Read the article in La Tribune (in French)