The atomisation of insurance

The insurance sector is on the cusp of fundamental change. And at the heart of that change is the sector’s use of data, primarily for the underwriting of the risks we present. Consumers are told that this will allow insurers to offer them ever more personalised products and services. There’s a lot to say for that, but like many things, there are two sides to it. In this post, I want to explore the implications of what I’m calling the atomisation of insurance.

A powerful argument for ever more personalised underwriting lies in what insurance people refer to as adverse selection. This says that an underwriter whose premiums are based more on a pooled rate than a personalised rate will find high risks being attracted to her portfolio and low risks leaving her portfolio, relative to a competing underwriter offering a more personalised rate. This is seen as only fair and it is then seen as just as fair to offer that little bit more personalisation.

The Pressure to Personalise

As opportunities to enhance the personalisation of your underwriting increase ( from all sorts of big data that suppliers are offering), then so do pressures to personalise your rates that little bit more. And ultimately, this becomes a never ending process, until of course, you end up with one-to-one underwriting. Personalisation becomes individualisation: a unique premium rate for each policyholder

Yet even then, we’re still looking at this through a traditional lens. Adverse selection would still be taking place along the dimension of time. Why individualise a person’s premium while still offering them an annual renewal? This then would drive premium adjustments towards a monthly and then weekly basis. And who’s then to argue against daily adjustments? Why not hourly? After all, some would see it as unfair not to. The internal logic of adverse selection then seems to become unstoppable, especially (and here’s the vital link) in a big data world in which firms have the opportunity to do something about it.

Think of telematics and the constant stream of driving data flowing from that black box in your car to the motor underwriter. Why add up its premium consequences in chunks rather than bytes? Delay seems almost like a historic anachronism.

Unfettered Access to Data?

Add to this an argument developing in insurance circles that anyone who doesn’t give underwriters unfettered access to their data must of course be a higher risk and so warrant a higher premium. So there are predictions of the insurance market splitting, into those who free up their data and get individualised premiums, and those who hold back on their data (or don’t generate enough) being automatically charged on the assumption that they are a higher risk. Sounds like a vulnerability and consent nightmare.

These sorts of developments can only be fulfilled in an era of machine driven underwriting. Gone will be a human touch to anything other than a minute handful of policies. And even then, that handful of policies may just be offered a renewal price designed to move them off that insurer’s books. After all (goes the argument), why incur expense that low risk policyholders will have to pick up the tab for?

This is not something unique to insurance. Many business sectors (and particularly financial ones) are on a micro-temporal trend, moving from the macro to the micro, even to the nano. While change seems inevitable, It could nevertheless have consequences for insurance that are unique

A New Insurance Landscape

So what sort of market will emerge out of what I’m calling the atomisation of insurance? Here are four features of the insurance landscape of the future.

1. underwriting will be increasingly based upon our daily experiences – life events happening in our homes, on our streets, at our work. And weaving these events together will be the underwriter’s analysis of the emotions and sentiments that guide our responses in various contexts. We’re already being categorised in this way – it’s the underwriting that just needs to catch up. Is the public ready for this type of relationship with their insurer?

2. One consequence of such a relationship will be how our daily experiences and sentiments are then influenced by this new insurance landscape. Behaviours will be modified – that’s inevitable. Some of those changes will be steps forward – safer drivers for example. Others will be steps back – less adventure, fewer behaviours outside of the norm. And less risk-taking as well – forget about any more people like Marie Curie. Some refer to this as the ‘chilling hand’ and talk about insurance morphing into a form of financial surveillance. Yet does this pattern stop there? Is not insurance then set on morphing into something that, on the deck of the USS Starship Enterprise, First Officer Spock would have referred to as “it’s insurance, but not as we know it.”

3. If adverse selection is seen as protecting lower risk consumers from higher prices, is it not also exposing them to greater pricing volatility? After all, lower risk isn’t ‘no risk’, so when the unfortunate does occur, the inevitable repercussions of individualised, machine driven underwriting will be sharp shifts in premium. The public are likely to find such volatility to be more disabling than the adverse selection against which they’re being protected. Of course such volatility will be experienced by the few, but then, like it or not, it only takes a few stories to influence opinions. So could the somewhat perverse outcome of ever greater levels of personalisation, be consumers experiencing ever greater levels of exposure? Micro pricing reintroduces macro exposures – odd, if you think about what insurance was invented for in the first place.Is insurance in danger of replacing risk smoothing with individualised ‘micro-crisis’ events?

4. Those who hold most data, and draw most insight from it, are seen to hold most power in a market. And while this can be a powerful force for innovation, it can also have side effects, most notably around the way in which that insight is interpreted. Data is generating a new language of significance, about what is reasonable, about what is typical, about what is fair. The great danger for insurance is that this new language will be formed within the confines of technical underwriting and data analysis – places with, we all have to admit, neither the diversity nor life experiences of the wider public. So questions such as about the the reasonableness of access to data, of adequate forms of consent, will be answered more from positions of power than positions of representation. Now, insurers have called for the public to trust them on this. And indeed, that trust may have to be renegotiated, but perhaps given recent surveys, insurers will have to work harder to get the public to sit down at that negotiating table in the first place. Could this be the most fundamental challenge facing ‘governance and insurance’?

So where do we go from here? Research – absolutely. Inclusive debate – double absolutely. Will it happen – that only gets a question mark at the moment.  Is it now time then for the insuretech debate to enter a new maturity?