We know that for UK insurers, price walking is on its way out. Yet it’s a story that’s far from over, for an ethically more charged issue still remains. The ethical risk that settlement walking represents is very real and far greater than price walking. This is what I meant at the beginning of this year when I talked about 2021 being a year in which insurers have to learn lessons, and particularly around claims. If settlement walking is not on a Chief Risk Officer’s radar, then the learning curve for their firm could be reputationally stressful.
Settlement walking is sometimes referred to as claims optimisation or street settlements. It revolves around settlement strategies being tweaked in to take account of two things: a claimant’s ‘willingness to accept’ and a claimant’s ‘willingness to complain’.
Two Types of Settlement Walking
With ‘willingness to accept’, the insurer uses data about the claimant to assess whether they would accept less than full settlement. Various tactics are used, such as the credit profile of the claimant, which points to their interest in a quick but reduced cash settlement. After all, it’s been argued to me, if the claimant is willing to accept less, then why not offer it to them.
With ‘willingness to complain, the insurer would, for example, track 1000 high volume claims of mid to low value. Settlement offers would be tracked and if a batch of slightly reduced offers didn’t trigger negative feedback, future offers would start at that slightly reduced level. The next batch would repeat this process, and so on until negative feedback like complaints start to rise. At that point, the slightly more reduced level of settlement is moved up a little, to reduce inbound complaints. The result for this insurer is a lower claims outlay, but with little discernible impact on complaints.
Now, some of you might be thinking: who would actually be doing that?! Isn’t this all a bit of conjecture? Well, look at a 2019 report from the Bank of England and the Financial Conduct Authority on Machine Learning in UK Financial Services. It confirms that 83% of UK insurers are using machine learning in claims management, in two main ways.
Firstly, ML applications are being used to analyse photos and unstructured data sources for claims cost prediction and total loss prediction. Secondly, ML applications are using predictive analytics to target claims that have a high likelihood of customer dissatisfaction or complaint, in which case they are flagged so a human can monitor the claim and intervene if required.
Now, we can see that the BoE/FCA report says two things. Firstly, that insurers are using ML applications to predict which claims might trigger a complaint. And secondly, they’re doing so in order to flag such cases for human intervention. So while the first thing demonstrates that systems have been designed to predict a propensity to complain, the second thing says that it’s being used to help avoid those complaints.
It's Not New
What is the problem then, you may ask! Surely that is a good thing. And it is, but then I think back to a 2014 FCA thematic review on household and travel claims. They found best practice guidance being used in a claims function that stated that claims were to be settled at below the correct amount if they felt the claimant would not complain.
So what we have then is an analogue era use of ‘settlement walking’ based on ‘willingness to complain’, and a digital era capability to do the same thing with machine learning. The sixty four thousand dollar question therefore is: is it actually happening?
I would put that question to the people best able to answer it: the heads of claims in UK insurers. Yet I would tweak it slightly: please show me the evidence to confirm that settlement walking is not happening in your firm.
Why seek evidence? For two reasons. Firstly, we are well past the ‘tell me’ approach to governance. Indeed, we’re coming out of the ‘show me’ approach and moving into the ‘prove to me’ approach. Secondly, the reputational impact of settlement walking is far greater than that of price walking. That is because while price walking happens within what is essentially a one-to-many relationship, settlement walking happens within a one-to-one relationship, with a ton of case law behind it.
This points to settlement walking, as I said at the start, being an ethical risk that is very much worth putting on a chief risk officer’s radar. Let’s assume then that it has gone onto that risk radar. What then?
There are a myriad of steps that need to be taken to track this particular reputational exposure. They stretch from performance management, to system design and testing, to due diligence, to system controls and supervision. What perhaps matters most though is an independent, informed yet critical eye. Someone who is able to see beyond the ‘we would never do something like that’ position.
No one within the market saw the unfairness in price walking, until a harsh light was shone on the almost universal practice from outside of the market. There are clearly some who haven’t seen any problems with ‘settlement walking’ through ‘willingness to complain’. Let’s hope historic doesn’t need to be repeated.