The words of the day after: Risks

Blog: Capgemini CTO Blog

The COVID-19 situation is shaking up our economies. For France, the estimated impact is at least 5 points of our GDP, i.e. around €120 billion. More than numbers, the crisis is a calling to question the existing development models and their fundamentals. Although the risks of a pandemic were theoretically identified and qualified, our globalized and inter-dependent economies were not prepared for a large-scale lockdown.

“It is impossible to assess the medium and long-term impacts: there is no relevant precedent, and demand and supply are both affected, which is unheard of.”

Governments have all announced massive and widespread support plans. While the shape of the post-COVID world is under debate – similar or fundamentally different – the probability of a resurgence of this or another virus cannot be ruled out, and with it, the possible re-occurrence of lockdowns, possibly shorter and more localized.

While tomorrow’s world will have to prioritize the treatment of causes (vaccine, medical treatments, more effective containment strategies, etc.), our economies will have to incorporate this new risk of lockdown and minimize its effects. Therefore, the impact of the current crisis have to be managed and systems put in place to cushion future potential occurrences, without just settling for tried and tested recipes. These approaches have demonstrated their limits faced with the complexity of today’s world. We must prepare more robust mechanisms faced with “unknown unknowns”, with an evidence-based and iterative experimentation approach (“Building a culture of experimentation”, Harvard Business Review, Apr-March, 2020).

Decisive role for Banking and Insurance: Risk management to be reconsidered amid pandemic and lockdown.

The economic management of this crisis is becoming imperative and our tools mustadapt. Banks and insurance companies are particularly critical in this approach.

First, due to their respective roles in society: Bankers and insurers must provide the tools to finance and protect against risk, which are fundamental pillars on which to build an economic future. Further, in the current crisis, the Government considers banks to be essential for the continuity of the economy and they are directly involved in the national solidarity effort (accelerated loan procedures, free deferral of repayments, overdraft facilities, and much more).

Secondly, because risk management and control are key to their business. For bankers, there is a default risk in the management of their assets, and for insurers, an underwriting impact. In view of the systemic nature of a large portion of these actors, any misjudgement of the risk can take on major proportions, like during the 2008 crisis. These two industries are also subject to scrupulous control by regulators (Basel, Solvency, IFRS, etc.). They are therefore limited in their room for manoeuvre and support cannot be provided at just any price. Banks in Europe have already called for a relaxation of prudential rules so that they can play their role in the current crisis, and in France, the Conseil de Stabilité Financière has authorized French banks to reduce their equity buffer to zero.

Bankers and insurers nonetheless face two different short-term situations:

As mentioned above, banks are directly involved in managing the crisis. Without any cynicism, this crisis is also the opportunity for them to play their role to the full, in a different register to that of 2008. Economic support through loans puts banks on the front line and increases their cost of risk. Whereas the latter has significantly reduced since the 2008 crisis (0.11% of the banking balance sheet in 2018, i.e. around €8 billion), it could go up again, weighing on profitability. Banks are therefore walking a fine line between solidarity and safeguarding the economic model. Initial estimates in China record a potential doubling of “bad loans” to $1,500 billion, i.e. an increase of 750 billion compared to the €300 billion guarantee proposed by the State in France.
Insurers are in a more complicated situation. Epidemic risk and the lockdown generated by this are generally not covered, notably when it comes to operating losses. Some guarantees, however, are directly impacted by the crisis, one way or another: increase of Death claims but reduction of Vehicle claims during the lockdown, for example. As far as society is concerned, the position of insurers is delicate and the expectations of a contribution to the national solidarity effort are pressing. Insurers announced support measures very early on (maintenance of guarantees in the event of payment difficulties, suspension of disputes, the contribution of €200 million to the solidarity fund, to name a few) and one even refunded a share of Vehicle premiums due to a drop in the claims received.
Finally, insurers and bankers are facing a challenge of operational efficiency to manage the increase in flows with sometimes reduced teams.

How to integrate this new risk?

Insurers cannot exclude epidemic risk and its consequences, in particular, during a lockdown. Doing so would be a failure of their duty of protection and the necessary contribution to the future of our economy. Like the risk of natural disaster, regulated in solidarity with the State[1], or guarantees against terrorist attacks via a dedicated fund[2], the risk of lockdown could also be the subject of a three-way discussion between civil society, the State and insurers, in order to define the fundamentals of coverage. Insurers have made a commitment to reflect with the State on this model for the future.

In future negotiations, the ability to establish financial issues proactively and objectively will be decisive.

Two subjects thus emerge:

The ability to adapt actuarial models to the risks covered. These technical impacts are generally absorbed by the models, especially the case of excess mortality taken into account by Solvency II. In parallel, some property uses will be reduced, significantly transforming some underwriting profits, particularly in property and casualty insurance (60% of the P&C insurance market).
The cover of operating losses resulting from the lockdown and therefore “without claims”. Although the demand is real and strongly relayed, actuarial models would only allow this risk to be considered on “limited and non-systemic³” perimeters. But, the systemic aspect is the defining characteristic of the impact of an epidemic. The polemic around the World Bank’s pandemic bonds reflects the extreme complexity of the models to be implemented to cover these risks and questions their affordability. The massive use of AI presents a new avenue but is now confronted with the lack of historical data. However, it’s a big step and one that will require State support, as reminded by Jean-Laurent Granier, CEO of Generali and Vice-President of the FFA, insofar as the €50 billion of estimated losses corresponds to the total equity of French property and casualty insurers.

“Bankers will also have to introduce the risk of lockdown in their credit scoring models, at the risk of seeing their results considerably increase, with repercussions on the cost and the impact on the actual economy.”

Once again, discussions will surely take place with the regulators and insurers due to the inter-dependence of the risks.

Intelligent use of data is critical for refining scoring models considering an industry’s or an individual’s sensitivity to the risk of lockdown. In this analysis, the ability to provide nuance (intelligence) will be vital. The business sector, the ability to work remotely, financial status, etc., are all criteria that need to be cross-referenced in order to qualify this new risk and integrate it into credit risk. Like the analysis models of the KYC file, the application of an artificial intelligence engine is a fundamental accelerator. In this case, it will be necessary to cross-reference:

Existing client data
Public data on the global situation by sector and activity type
Any additional client data

This quantitative analysis will have to be confronted with political reality and then shared with civil society and the State. Using only statistical analyses carries the risk of only financing pharmacies and not a single restaurant.

“For bankers and insurers, data modelling work and the development of AI algorithms will now be the key factors for an efficient relaunch.”

AI also at the service of flows and the fight against fraud and vulnerabilities

Back-offices will deal with document flows to be processed thanks to OCR (image processing) and NLP (word processing) tools that are now robust and versatile. These approaches require less data to train the algorithms, benefiting from the transfer learning technique that has proven itself in recent years (The Economist, Artificial Intelligence, 13th May 2017).

These automatic document processes can be enriched with “learning” fraud detection tools that do not require the implementation of a series of rules. They are based on error detection approaches used for example in spam filtering, completed with explainable meta-algorithms that filter out false positives (European Banking Authority Report on Big Data and Advanced Analytics Jan 2020).

Further, banks that quickly apply algorithms and train them on data from the current crisis will develop a comparative advantage in terms of service, credit risk control and a head start in the discussions about the future.

For insurers, the use of this new data will have to optimize the actuarial models within the context of the creation of products adapted to the new risk mapping with which they are faced.

The two industries will also have to be major players in dealing with the most vulnerable members of our society. AI opens real prospects to anticipate and remedy them.

This article is an English adaptation of a post initially created in French.

This blog is co-authored by Moez Draief

[1] Natural disasters: covered in vehicle and home insurance contracts based on a standardised percentage of the damage premium and with specific reinsurance by the state
[2] Terrorist attacks: indemnity by the FGTI (guarantee fund for victims of terrorism and other offences), funded by a lump sum on each property insurance policy
[3] Hervé Marzal, director of Gras Savoye Risk Consulting – L’Argus de l’Assurance.