The researchers, Pierre Francois and Théo Voldaire of Sciences Po in Paris, studied two leading French motor insurers. Both firms were composites and amongst the top ten insurers in the motor market. One was a mutual focussed on France and the other was shareholder with greater geographical diversity. Four telematics offerings were studied, two from each firm.
Their research producing four main findings:
- moral and political factors were only secondary to how France’s telematics experience played out;
- there were organisational and cultural barriers hampering the adoption of telematics;
- the different financial and governance structures (mutual and shareholder) were not significant factors;
- the main factor was both firms adopting an experimental strategy and found that “in the short to medium term at least, the use of big data to rate car insurance is not profitable.”
So does this mean that telematics, and personalised underwriting more generally, is just over-hyped? This is not the only study to find that big data underwriting has not been taking over the market at pace, as some predicted. And a good many start-ups using the big data model have had some very challenging years of late, if not worse.
The landscape is more nuanced, and to a certain degree, these two (unnamed) French insurers knew that. Their telematics strategy was experimental because they wanted to learn exactly how and where it was most nuanced. Both knew they would loose money on this initial offerings of telematics products, and both were prepared to withdraw them, as has happened in three out of the four products.
This is not a Failure
So this is not a failure of telematics in French motor insurance. It looks more like a ‘dipping the toe in the water and taking it out again’ approach. Clearly, if you’re a start-up firm, there is no dipping of toes. You have to go into the market with at least one foot in the water, probably both feet in at the same time. Hence their mixed fortunes.
Personalisation is not going to go away. That’s because it is neither a technological thing nor an insurance thing. It is a social trend to which businesses are adapting and learning how to apply technologies to remain relevant and profitable. Nothing new there then.
What these two French insurers learnt was that the use of personalised underwriting technologies were not an immediate or natural fit for their motor market. And this is despite telematics having been seen by many leading insurance markets as the most obvious ‘best starting point’ to make use of personalised underwriting technologies.
The obvious question then is whether this situation will change, or is telematics always going to struggle? The researchers see this as largely a question of market dynamics. Telematics has become established in Italy, so why not France? That’s what we’ll look at now.
The researchers weighed up the pricing dynamics in relation to four types of drivers...
- there are drivers with good telematics scores and no claims. Their premiums are low and they are not very profitable for insurers.
- there are drivers with good telematics scores but large or frequent claims. Their premiums can go up only by so much (the ‘no discount’ rate or what French law allows). They may even earn discounts from their telematics score. Their claims make them very unprofitable to the insurer.
- there are drivers with poor telematics scores, but no claims. Their premium goes up to the no-discount rate or to what French law allows. They are very profitable to the insurer.
- there are drivers with poor telematics scores and large or frequent claims. Here the rate increases cannot balance the level of claims and they are very unprofitable to the insurer.
There are two key points being made here. The first is that telematics scores and claims outcomes do not yet tally. Either the portfolio is too small, the time period too short or the scoring methodology not synchronised enough.
The second is that a discount approach to rewarding or penalising driving behaviour limits underwriting responses. The insurer’s telematics hands are tied so long as the score does little more than adjust the actuarial premium.
There are two areas which I think the researchers could have explored further. Firstly, with regard to the motor portfolio’s reinsurance arrangements. These would undoubtedly cover both telematics and non-telematics products under the one treaty and would, in the main, move a fair chunk of the large injury losses out of the primary insurer’s results.
And the second is that for drivers with large or frequent claims (whatever their telematics score), the insurer still has the scope to raise the actuarial premium, off which the telematics discount is then applied. They then have more scope to make such drivers less unprofitable than the researchers suggest.
What these two factors do is make telematics less of a ‘pricing disaster’ than the researchers postulate. What I think the researchers found was not so much a pricing strategy challenge, but a ‘model integration’ challenge. It wasn’t easy to open up your main, very stable, reliable pricing model to something with a lot of unknown or underexperienced data points / rating factors. The risks were high, while the returns were low.
Has the Revolution Happened?
The researchers end by questioning whether personalised pricing has really gained traction. In terms of the French motor market, it hasn’t. In terms of the Italian motor market, it has because premiums, especially in the south, are much much higher than in France.
Does this mean then that personalised pricing can only take off in markets with high average premiums? I don’t think this is the case and I’ll illustrate this with Flood Re, the vehicle for covering high flood risk areas in the UK. Consider these aspects of both cases:
- telematics is usually focussed on a small market (young drivers), while flood insurance was a large market with wide demographics;
- telematics introduces new rating factors, while flood insurance doesn’t;
- telematics is a mix of both cat and working risk, while flood insurance is cat only;
- telematics is a discount product, while flood insurance is a penalty product;
- telematics is a loss leader product, while flood insurance isn’t;
- telematics relies on behavioural risk management (driving behaviour) while flood insurance has very little of this.
In essence, the two risks were very different, most crucially in that personalised flood pricing can be handled within existing pricing models and systems. That is why premiums for flood insurance in the UK grew exponentially in the three to four years in the run up to the launch of Flood Re in 2016. Data about individuals properties could be utilised immediately within traditional pricing models.
And in a noticeable difference to the French telematics experience, it was moral and political factors that were then triggered, leading to the creation of a single national pool for high flood risks.
So I would say that we already have personalised pricing but its pace and scope are being determined by a range of factors. Looking a year or two ahead, what we saw with flood risk will soon be seen with escape of water risk. It’s all the talk of the sector at the moment and like flood underwriting, requires no significant changes to the pricing model.
Price is Not the Only Lens
One gets a strong impression that the French motor insurers were keen to launch these test products, but were mainly watching how the market and their competitors would react. They didn’t seem to put much energy into listening to how customers were experiencing these products.
This is surprising because the French market seems to put a lot of store on customer loyalty. As with the recent research into telematics in the Italian motor market (more here), it is the manufacturer experience being explored, not the customer one. How much of an opportunity was lost by not listening to the latter? Or to put it another way, how much of a research opportunity still remains?
And it is a very big opportunity. The danger is that the focus is on motor insurers, who see telematics through a pricing lens. What if consumers see telematics through a different lens? I think there are differences.
Proximity v Intimacy
We know that many people think they’re a better driver than they really are. What telematics offers drivers is insight into how honest they’re being with themselves. So I believe there’s an appetite amongst many consumers for feedback on the quality of their driving and the use of cues about how to improve. I say ‘many consumers’ because of course, there will always be some who don’t want their views on their own driving to be challenged, even by a mobile phone app.
In a 80/20 world however, I think 80% would be interested and 20% not interested. And if you can, through telematics dashboards et al, engage 80% of drivers in ways that could improve their driving, then the impact on accidents, and thus both property and injury claims, will be noticeable.
So why hasn’t this happened? I think it hasn’t happened because insurers adopted telematics within a ‘proximity’ digital strategy – one based upon using data to get closer and closer to the consumer, whether they liked it or not. If instead they had used a ‘intimacy’ digital strategy – one based upon doing things that causes the consumer to want to get closer and closer to the insurer – then telematics would have focussed less on price and more on claims reduction through better driving.
The proximity digital strategy approach is more short to middle term than the intimacy digital strategy, so perhaps that was why the market adopted telematics in the way that it did. Yet, reading Francois and Théo’s paper, I couldn’t help but think that these two insurers may have started down a short to medium term road, but in the end, they found that it was a lot more long term than they anticipated.
To Sum Up
It’s a valuable piece of research that tells us a lot about the complicatedness of innovation. It lacks the consumer’s voice, and this is where I hope further research will be undertaken.
It tells us that this or that market cannot be taken for granted when technology initiatives are being planned. Whether by country or class of business, each has its own dynamics that need to be accounted for.
For me, the motor market is always interesting – after all, I was once head of insurance at Europe’s biggest motor fleet. But at the same time, I believe the one market that will be the anvil upon which personalised underwriting is truly forged, will be life and health. All else is preparation.
That is why the EU, in its AI Act, has classified life and health insurance as the one market that is high risk. For European insurers, and to a large degree worldwide insurers too, how the EU on the one hand, and life and health insurers on the other hand, approach putting into operation that high risk assessment – this will be the crucible in which the future of personalisation is shaped.