Historically, insurance contracts have suffered from information asymmetry, with the proposer knowing more about the risk they are presenting than the insurer. The latter asked questions of the former through some form of proposal, in order to reduce that asymmetry and so be more confident about how they were pricing the risk.
In today’s world of digital insurance, it is possible that the information asymmetry could be reversed, through the use of all that data and analytics. The proposer knows pretty much the same as they have always known about the risk. However, the insurer has collected a lot of data points around the risk and uses analytics to gain new insights into granular risk patterns upon which to base a premium.
Why Bother with Questions?
This direction of which this development is taking insurance is exemplified by Aviva’s ‘Ask Me Never’ product, which relied simply upon the proposer identifying themselves. Nothing more was needed from them, for the insurer could get what they needed from their big data loch and provide a quote based upon that. And perhaps even that identifying bit could be done away with, by the use of image and voice analytics.
What is happening here then is a transition from the proposer knowing more about the risk, to the insurer knowing more about the risk. Information asymmetry is reversed. We’ve replaced adverse selection with inverse selection.
Sure, proposers are spared having to answer lots of questions, but instead, now find that they’re in what might be called a sellers’ market. The insurer will offer you a quote only if their new informational advantage signals that they can write your business on a profitable basis. If it doesn’t signal that, then they’ll either no quote you, quote you a ‘go away price’ or signpost you to some other provider.
And if that other provider has a similar level of informational advantage over the proposer, the proposer will be forced to play a game of merry-go-round signposting. There’s evidence of this happening in some segments of the travel insurance market.
Now some of you may say that this is less of a problem than it appears, for all of the data being collected by the insurer comes from the proposer in the first place. That’s true, but the key difference between the two parties is the analytics that allows the insurer to draw risk insights from all that data. So this swing from adverse to inverse selection is down less to the data and more to the analytics.
Goodbye Insurance?
The problem that then comes to the fore however is when you scale this up, on both buyer and seller side, to the level of the overall market. The more data (or the cleverer the analytics) the insurer has, the more likely that they’ll become more selective in what they write. The result is a market that atomises, with insurers more and more inclined to take on less and less risk. This in turn leads to an implosion of provision, with insurance becoming less about risk transfer and more about savings and risk management. As the saying here in the UK sort of goes, will the last insurer leaving the market please switch off the light as they go out.
Just as much as insurers would worry about adverse selection, perhaps the insurance buying public should now worry about inverse selection. There’s something rather ironic in all this. That worry of insurers about adverse selection was a key argument for their accumulation of data and analytics, as tools for tackling it. Yet in so doing, it has led to the emergence of something just as significant and concerning for their customers. Is that what digital insurance ends up delivering for the public?
And while insurers often talk about data and analytics heralding in a new transparency and openness about the market, from the customer’s perspective, that’s of little use if the market is progressively backing out of reach.
What Could Go Wrong?
The picture I’ve painted above is rather smooth, when in reality, there will be bumps and twists in all this. The most obvious bumps are, firstly that the data may be incomplete and/or be misinterpreted. And then the analytics may generate insight that is, let’s say, not very good. So insurers will not always get their pricing decisions right.
Another bump would come from the analytics relying on statistical significance, when in reality, the policyholder relies on their knowledge and experience of real life. So the underwriting decision system may receive the wrong risk signal and charge an inappropriate price, either too high or too low.
The public will experience these bumps by being closer to the actuality of the risk. What can then results is a lack of faith in the market. Thoughts like ‘do they actually know what they’re doing’ arise. I’ve seen one insurer price a risk through one channel at a tenth of what it was wanting through another channel.
What to Do About It
What can be done about it then? I think a good start would be to rethink the language. Inverse selection is so named to be in relation to adverse selection. The language of ‘they are getting one over me’, either which way, is not conducive to evolving a market into a form that works for both insurer and insured.
Then I think insurers need to think more carefully about their narrative on adverse selection. There needs to be greater recognition of the informational switch that is now happening. What we might then need is simply ‘selection’, acting as it does in both ways.
About twenty years ago, I did some research for a university course on utmost good faith and how the responsibilities it created actually acted both ways. The insurer was just as much bound by it as the policyholder, yet the narrative was always about the latter’s responsibilities. Something similar now seems be happening around selection.
It’s time for a more open and honest look at selection.
