Is the insurance industry using too much data for underwriting?

The trend around insurers offering personalised pricing to insurance customers can work to lower premiums for some demographics - but there are major downsides to consider.

For example, the more policyholders who receive personalised premiums, the more this dilutes the advantages of pooling risk, which is core to the concept of insurance.

There are also ethical issues. For instance, unless an individual’s job involves driving, why should their occupation affect their motor insurance premium?

Research published by price comparison website GoCompare in March 2022 revealed librarians and barbers to be among the top 10 professions paying the most for motor insurance.

This could be because insurer data finds that these professionals often have to commute further as they can’t afford accommodation near their place of work.

Meanwhile, a February 2020 report commissioned by the ABI and conducted by research institute Britainthinks, titled The price of accuracy: Consumer attitudes to data and insurance, showed a divide in consumer opinion around data.

Just under two-thirds (64%) of the report’s respondents would prefer all insurance customers to pay for their cover exactly according to their level of risk – even if this makes insurance unaffordable for some people. However, 36% of respondents took the opposite view.

James Daley, managing director of consumer group and consultancy Fairer Finance, is highly supportive of this latter minority.

He complained that there are few rules and little transparency regarding what is behind peoples’ pricing – leading to some decidedly unfair outcomes, in his opinion.

He told Insurance Times: “Why are unemployed people being charged more than stay-at-home parents? And why are drivers involved in non-fault accidents being penalised with premium increases?

“It’s time for us to have a national debate about all this and draw up some rules. Our insurance industry has lost its heart, like a robot, so no wonder it isn’t trusted.

“[The industry] hides behind the idea that the way it underwrites is its secret resource, but I think it’s one of its greatest flaws and will drive trust down further.”

The fact that the UK government and regulators have been less focused on data laws than their European Union (EU) and US counterparts hasn’t helped either.

There is little relevant UK legislation applicable for monitoring data used for underwriting, other than the Equality Act 2010, which protects against using data to discriminate, and the General Data Protection Regulation (GDPR), effective from January 2021 in the UK following the EU’s May 2018 version, that aims to safeguard consumer consent - with questionable effectiveness.

“You have to give your consent for data to be used,” continued Daley. “But in practice, everyone just clicks [yes to] everything online as they are overwhelmed with requests.”

No right or wrong

Ben Howarth, climate change and open data policy manager at the ABI, believes the issue is less about data being off-limits and more about the importance of firms having robust governance to check they have clear processes for using data effectively.

However, insurers already claim to have this governance structure in place. Axa UK, for example, has a pricing and underwriting ethics committee that meets regularly to discuss all new data prior to its use in modelling.

Howarth also argued that increased innovation around data use could enable insurers to make more accurate predictions rather than using current proxies.

Furthermore, he believes there is justification for current data practices based on the grounds of tradition.

He noted: “As far as I’m aware, non-fault accidents and occupations have been used in underwriting for a very long time and have not arisen because of increased data availability.”

This latter defence is curious, however. Just because something is standard practice does not necessarily make it acceptable.

While some underwriting practices cannot be considered right or wrong, opinion on them certainly varies.

Matt Carter, practice director for insurance specialty markets at consultancy Altus, said: “There shouldn’t be penalties for non-fault accidents. The technology and data exist for insurers to be able to disregard non-fault claims.”

But Matt Connell, director of policy and public affairs at the Chartered Insurance Institute (CII), has no problem with the 10% motor premium increase he received following his own non-fault accident this January.

He explained: “I can see why the premium has gone up. The insurer will feel I frequent routes where accidents are more likely to happen. It says something about the riskiness of my driving environment, even if the accident isn’t my fault. It might not be my fault again next time.”

Clarifying data boundaries

It is hard not to support Daley’s call for an industry-wide debate to draw up some solid rules regarding what is permissible around the data used for underwriting. This kind of discussion could shed light on areas where it remains unclear what activity is actually occurring around data use.

Evan Moore, head of personal lines underwriting at Zurich, has heard anecdotal reports of other UK insurers using social media for underwriting.

Some industry commentators also surmised that loyalty card data is being used to inform underwriting - the refusal of Sainsbury’s or Tesco to comment on this subject when asked by Insurance Times creates more suspicion than clarity around this.

Once greater transparency around underwriting data has been achieved, an educational piece for consumers should be a high priority as well.

Alarmingly, 70% of general insurance consumers featured in the aforementioned ABI and Britainthinks report incorrectly thought that gender is taken into account when pricing insurance. However, this was outlawed by the regulator back in 2012.

There will always be winners and losers from personalised pricing. But, with technology having outpaced legislation, the insurance industry badly needs to establish what is ethically acceptable and how far it can go before the advantages of cross-subsidy, where one group of consumers pays a higher amount so that the price paid by another group can be reduced, are lost.

Particularly worthy of investigation is an idea previously discussed and mooted by the Institute and Faculty of Actuaries’ The hidden risks of being poor: The poverty premium in insurance report, published in September 2021.

This mentioned using the concept of Flood Re to establish a Postcode Re, to help those unable to afford compulsory types of insurance.