Why insurers need to review their use of credit data

The US auto insurer Root recently announced that it will remove credit scores from its insurance pricing by 2025. The reason – because of the “unfair, discriminatory biases” that it introduces. It’s a wise move, although it still comes with a host of questions attached. UK insurers would do well to think carefully about credit data as well, for it rings a number of warning bells with the regulator.

Some context first. The regulators of all but three US states allow credit scores to be used in pricing models. California, Massachusetts and Hawaii are those three states who won’t approve any rates from a pricing model that uses credit scores. And their reason is the same as Root’s – its unfairly discriminatory. So the US is far from being uniform on this issue. This means that auto insurers aspiring for market greatness have to run two pricing models, one for big states like California and another for those many states who allow it.

A Tale of Two Pricing Models

Now, Root is licensed in many states, including California, so they must have within their pricing model, an existing capacity to price both with, and without, the use of credit scores. This makes their planned date of 2025 for the elimination of credit scores, on first glance, a puzzling one. More on that in a minute.

The second puzzle is around their focus on insurance pricing. No mention is made of the use of credit scores in risk selection, policy make-up, counter fraud or claims assessment. In the US and the UK, credit scores are widely used across all of these activities. This is not me being pernickety. This is about clarity and completeness.

Remember as well that it is relatively easy, in data science terms, to generate proxies both big and small that can stand in place of the actual credit score. So an obvious question for Root would be whether they are also eliminating the capacity for their various underwriting, claims and counter fraud models to generate outputs that replicate the essence of credit scores. If not, then what’s the point?

The Point is Marketing

The point, I believe, of Root’s decision can be found in the insurance function we haven’t mentioned yet – marketing. They’ve picked up on research showing that credit scores can produce unfair and discriminatory outcomes, and that it is more than unpopular with consumers. And they’ve sensed that this ‘writing on the wall’ points to it being a good time to make a public relations play for attention and differentiation. So while they will already be running a pricing model in California that doesn’t use credit scores, their portrayal of its elimination across the board as a big, awesome challenge is largely an attempt to capture some marketing high ground.

So what ‘writing is on the wall’ for the use of credit scores in UK insurance? Well, it’s a substantial list. The current regulatory interest in the fairness of retail pricing is an obvious avenue down which the use of credit scores can be scrutinised. Then there’s the FCA’s vulnerability programme, with its obvious associations with creditworthiness. Looking a bit further ahead, there’s the market studies on data ethics and credit information. What this adds up to is the use of credit scores in UK insurance is an ethical issue both active and growing. Time to pay it some attention perhaps?

Credit Data is Not a New Issue

It’s worth remembering how the UK regulator approaches credit scores. There is no explicit nod within the rule book for its use – just an absence of any ‘disallowed’ signal. Yet the way in which credit data is used in pricing and all other functions does explicitly fall within the Principles for Businesses, and all of the rule sets and programmes associated with them.

This means that UK insurance executives should not see their use of credit data as an issue only now emerging onto their risk radar. It is, and always has been, a current issue, for which insurers should be able to provide evidence that it does not contravene those all important Principles for Businesses. How many insurers have that evidence to hand? Given existing corporate and individual accountability under the Senior Managers and Certification Regime, it would be wise to ensure that it’s up-to-date and solidly critiqued.

Solid Evidence

So what would ‘solidly critiqued’ evidence look like? Here are some thoughts…

• It should be gathered and analysed by a third party who has not previously been involved in your pricing strategy;
• It should cover all functions – underwriting, claims, counter fraud and marketing;
• Both data and analytics should be tested, with attention paid to both direct and proxy data.
• It needs to be undertaken within a ‘critical friend’ remit;
• It should be informed by the work of civil society groups with an interest in credit scoring.

There is one last point to be emphasised here. Root’s decision to remove credit scores is a response to an issue they see as under-recognised at the moment. It won’t stay like that for long. And here in the UK, that’s certainly true – the FCA market review points to that. So the mood music around credit data is changing, and UK insurers need to tune into it. Just because they have been able to use such data, doesn’t mean they should continue to use it.