The Financial Ombudsman Service (FOS) has been weighing up the implications that big data will have for its work. And the implications are indeed significant. Unfortunately, the big data report commissioned by FOS misses the most significant implication of them all. So there’s a danger that claimants will find FOS struggling to arbitrate effectively around a structural change underway in insurance claims.
The Future Foundation’s report for FOS concentrates on two impacts that big data will have on its work:
1. big data will allow FOS to more easily identify patterns of detriment and so allow it to better arrange its resources to arbitrate them;
2. those patterns of defriment could also be identified by other providers and so open up the arbitration sector to new entrants offering disgruntled claimants new ways of seeking compensation.
These are certainly developments deserving of FOS’s attention, but overshadowing them is a change underway at the very heart of insurance claims.
Decision making in insurance claims is based, in very simple terms, upon firstly, the information the insurer has on record about the risk being insured and the person insuring it, and secondly, the information being supplied by the claimant about the loss they have incurred. Todate, both of these sets of information have been provided by the same person, initially when taking out or renewing the policy, and then at the point of claim. Differences between the two sets would raise questions about the validity of the claim, perhaps even about any fraudulent intent of the claimant.
Big data is changing this. Increasingly, the data about the risk being insured, and the person insuring it, is being drawn from an ever widening variety of sources. A few of those sources will relate to insurance, but the majority will be based on data collected in other circumstances, such as credit checks, health records, maps, shopping preferences and social media chatter, to name but a few.
What big data will do is allow the insurer to aggregate it all together to create an identify for the insured risk, the insuring person, upon which underwriting decisions will then be based. When a claim comes in, the picture built up by the insurer will then be compared with that presented by the claimant. So how will differences between the two be handled?
Let’s say that the insurers’ indentity for the policyholder contains characteristic X, but the policyholder has put on her claim form that she has characteristic Y. The difference is material. As a result, the insurer turns the claim down.
Yet what if the insurer’s information about characteristic X came from a datasource unrelated to insurance? The data may have been disclosed by the person under quite difference circumstances, perhaps just as a joke between friends on social media, or via an anonymised dataset that was subsequently re-identified by a data broker working for the insurer. The quality may be suspect, the provenance confused.
Yet from this could emerge a dispute about the validity of the claim, about the truthfullness of the claimant. How will FOS arbitrate this? I suspect that it will say to the insurer that the accuracy and completeness of anything other than what was disclosed directly to it by the policyholder is an assessment process of its own making and any risk from it must be for its own account, not the policyholder’s.
This could become an increasingly common occurrence. So big data will not only help FOS identify new patterns of detriment, but actually introduce one itself, perhaps, as the influence of big data spreads, the most significant and pervasive of them all.
An insurance sector flush with underwriting insight from all that rich big data may find this a difficult situation to swallow. Much better perhaps to address it now, through dialogue with the Financial Ombudsman.