A pattern is emerging across the market that points to a new type of underwriting. And it is pretty revolutionary, as well as controversial. It is being referred to as ‘underwriting at claim stage’. It fits wonderfully with the big data trend, as well as various user friendly initiatives, but fits uncomfortably with ethics and trust.
Underwriting at claim stage involves insurers only taking into consideration information material to their assessment of the risk at the point a claim is made. You can see the attractions for insurers. They are able to put aside those awkward questions until when they really matter, which is at the point of claim. The huge number of people who don’t claim needn’t be bothered with all that detail, all that hassle. It all seems so friendly, on paper.
All this is a natural outcome of the sector’s drive to utilise data and analytics to make the experience of engaging with insurance so much easier. Some insurers are seeking to eliminate the need to ask consumers any questions when giving them a quote. Instead, they want to rely on big data to draw the insight needed to underwrite the risk being presented.
What strikes me about this trend is this. If you don’t ask the consumer any questions when accepting their business, then it would be pretty difficult to accuse them of non-disclosure when vetting a subsequent claim. After all, the customer would not know what big data you used, or how you used it, to underwrite their policy in the first place.
Underwriting at claim stage (let’s call it UaCS) seeks to be the market’s way of getting round that problem. When that now customer submits that claim, databases of underwriting information are analysed for signs as to how the claim submitted matches with the risk (both policyholder and insured object) the insurer understood they were taking on at the time of inception.
So no matter that the insurer only asked for your name and address when taking out that policy. If they find things at the point of claim that would have troubled them at the point of underwriting, then onto the reject pile goes your claim.
Now every insurer will at this point be thinking that it’s very in order for them to make enquiries to validate the claim and claimant. It’s only fair that the honest customer doesn’t pay for the claims costs of dishonest claimants. And there’s a lot of mileage in that argument, but I’m not convinced there’s enough mileage in it to fully justify some uses of UaCS. I’ll explain why under three headings.
All parties to insurance contracts are bound by rules relating to representation. In the UK, these are set out in the Consumers Insurance (Disclosure and Representations) Act 2012. This act did away with the age old concept of utmost good faith. Interestingly, that old utmost good faith obligation, while invariably talked about in terms of ‘consumer to insurer’, was actually a two way obligation. It applied to ‘insurer to consumer’ representations as well.
Now, I’m not lawyer, but the 2012 Act reads as very much a ‘consumer to insurer’ piece of legislation. I recall from the time of its Parliamentary Committee stage that there was an attempt to include an obligation on insurers to make reasonable enquiries, but for reasons I can’t recall, that didn’t end up being incorporated into the eventual Act.
This matters because clearly if an insurer chooses not to ask a question, then its options for then relying on what it would otherwise have known are affected. One approach of course is for the insurance contract to be worded such that when you sign it, the consumer’s obligations are kept as wide as possible. How fair is this? It’s debatable, and certainly hardly trust inducing from the point of view of something on the proverbial Clapham omnibus. Yet with the digital habit of casually clicking on unread terms and conditions with more words than War and Peace, are we surprised?
What then happens is that these UaCS activities are conducted within a fraud investigation wrapper. This is not underwriting, but reasonable enquiries relating to possible misrepresentation by the claimant, so the argument goes. And again, it’s an argument that does have legs, but how far can you take it? If you ask few or no questions, then non-disclosure is different to when you ask what a price comparison website can put consumers through.
The problem that insurers don’t seem to recognise is just how much this systemically weakens their push to tackle fraud. With the sector using its own (and more relaxed) definitions of ‘fraud’ and ‘proven’, and applying fraud and UaCS processes under a sector wide regime overseen solely by insurance people, within an insurance organisation that is not subject to regulatory oversight, then opportunities for ‘scope creep’ seems an ever present risk. It happens – look at what the Korean market experienced several years back.
And the great danger from that scope creep is that with fraud being such a powerful label to stick on someone, the public could come to feel that it is too powerless to raise a challenge. Until that is, what an insurance fraud leader once described as his worst nightmare – the politically or media connected person who stops to think about why their claim has been turned down for misrepresentation. Then that power balance shifts away from insurers.
One last fraud point. A key database used by UK insurers is CUE – the Claims and Underwriting Exchange. At the underwriting stage, insurers subscribing to CUE can only run enquiries relating to the particular type of product being underwritten. With fraud investigations however, they can run enquiries across all CUE products. This changes the scope of their misrepresentation radar considerably. It makes the question of how insurers’ interpret fraud critical.
What permeates all of what we’ve covered so far is data and the analytics that pulls it all together. So an insurer asking few or no questions outwith of the consumer’s name and address is instead relying on the insight about that consumer that its analytics can draw from all its big data.
Yet consider what Professor Terras of Edinburgh University’s Futures Institute said at an Alan Turing Institute lecture earlier this week:
"All data is historical data: the product of a time, place, political, economic, technical, & social climate. If you are not considering why your data exists, and other data sets don’t, you are doing data science wrong”
What this means is that the data that a typical insurer holds on a typical consumer is not only incomplete, but out of context and out of date as well. Obviously, if you run underwriting checks at the point of claim, it is hugely out of date, to say the least. It’s underwriting on shifting sands. That’s why UaCS feels unfair both as a concept and as a practice, even after factoring in the importance of tackling fraud.
And putting data aside and looking at UaCS through only an analytics lens, the insurer is moving from one-to-one confirmation on specific points at the underwriting stage, to a mix of one-to-one data checks and one-to-many correlating enquiries at the claims stage. With a claim being such a one-to-one situation, the role of analytics in weighing up the validity of representations seems to stretch reasonableness that bit too far.
Now I know insurers will point out that policyholders are under a continuing obligation to inform them of changes in the risk. Yet if the data used to underwrite the risk is so vast (it’s big data) and so unknown to the policyholder (only asked for name and address), then it hardly seems a reasonable basis of response. How does UaCS and big data change the nature of that continuing obligation?
To wind up. Underwriting at claim stage is controversial, to say the least. If you read media reports about it, it just leaks trust in so many directions. The question that observers of the sector will be asking is this: what are insurers getting out of UaCS that is so big that they’re willing to run the accompanying reputational risk?
The answer of course is that they’re weighing up UaCS using a faulty pair of scales. The sector is experiencing a single minded push on data and analytics (there lies all the solutions), while being too insular on how it tackles fraud (only we know how to do it). The outcome, if insurers are not careful, will be a political and social kickback against practices like UaCS and fraud investigation that will take the sector by surprise. I’m seeing some initial signs of this already. The question for the sector is: are you willing to see them too?
Duncan is the founder of the Ethics and Insurance blog and the author of its many posts. He's a Chartered Insurance Practitioner, having worked 18 years in the UK market. As an adviser to many firms on ethics issues, as well as a regular conference speaker, he is one of the leading voices on ethics and insurance.