In the first of a series of articles, we look at the work the company has led surrounding the growing use of modelling in the insurance sector. Models are becoming ever-more sophisticated and prevalent. But what is the potential impact of this increased reliance on models? In this piece we ask the question: Why does it matter?
PayPal billionaire Elon Musk, thinks that rapid improvements being made to artificial intelligence are probably the “biggest existential threat” faced by the human race.
His longer-term vision, according to The Times “…is to enhance brain power to help humanity avoid subjugation at the hands of increasingly intelligent machines.”
His big vision is, as usual, eye-catching and headline-grabbing.
Another slightly less well-known figure has also been pondering the impact of technology and its impact on human behaviour – but this time its potential impact on the insurance industry.
Until a couple of years ago Oxford academic Anders Sandberg knew nothing about insurance. And even less about reinsurance.
He’s a neuroscientist who has spent much of his time researching the ethical and social issues of the relationship between human beings and machines.
In short, like Elon Musk, he spends time thinking about the future of humanity. Or more precisely, the future of insurance.
He is one of 30 academic researchers and industry practitioners, including insurers, catastrophe model vendors, brokers, regulators and consultants who have been pondering the ‘Systemic Risk of Modelling in Insurance’, or SRoM.
So far the group has produced a widely-read White Paper which looks in detail at the issue. It may not sound like your ideal bedtime read. It’s a big, complex topic. But an important one that is occupying many in the insurance sector.
Modelling – the use of software to automatically calculate risk or manage capital for example – is becoming increasingly prevalent.
The insurance and re-insurance sectors are becoming more and more reliant on modelling. But the work Anders and his colleagues are doing is asking an important question: What, if any, are the dangers of relying too much on computer models?
And more importantly if everyone is using the same models, is there a danger that everyone could make the same mistake and maybe cause a catastrophic failure of the whole market?
After all we came close to falling over the cliff edge with the banks back in 2008. It demonstrated that the excessive reliance on modelling wasn’t just an existential crisis. It was a very real one.
As Anders, senior fellow the Oxford Martin School (OMS) says in his introduction to the topic: “The term 'systemic risk' is often used when describing the Global Financial Crisis of 2007-2008.
“The catalyst for this financial meltdown was the bursting of the US housing bubble which peaked in 2004, causing the value of securities tied to the US housing market to plummet and damage financial institutions on a global scale.
“Core to the magnitude of the losses experienced by these events was the modelling assumptions upon which these securities were valued...”
In other words, because the bankers were all relying on the same models, when the models got it wrong the effects were devastating and reverberated across the banking sector.
This in turn created a ripple effect across the whole of society which is still being felt to this day.
So why does it matter in the insurance industry?
Systemic risk is increasingly relevant due to the growing level of globalisation, interconnectedness, and speed of business transactions in the world. MS Amlin and OMS have been working together on the project since the start of 2014, with the aim of researching and promoting a better understanding of SRoM across the insurance industry, to the benefit of clients and stakeholders.
It’s important for the industry because apart from helping economies to function, insurance has a social role to play.
Insurance is important for society as a whole. As the Chartered Institute of Insurance asked in a paper recently: “Imagine a world without insurance.”
Imagine the implications of not being able to insure your car or your house. You’d have to have considerable sums of money set aside in case of accidents or damage.
Or how about the companies who are investing in countries prone to natural disasters?
How would their economies grow without being able to insure against possible damage and financial loss?
So it’s important the industry as a whole takes the issue seriously and makes sure that our increasing reliance on computer models isn’t posing a risk to the system as a whole.
In our next article we will take a close look at what Anders and his colleagues have discovered during their research.