The next twelve months will see the pricing practices of UK insurers and intermediaries put under close scrutiny. The regulator is specifically interested in dual pricing in insurance and in its 2018/19 business plan, has confirmed that it will focus in on the household market. So what should insurance firms be on the lookout for, and how might they prepare for the spotlight about to be turned on them?
The FCA is concerned about how existing customers are being treated. Are they being overcharged for their cover in order to fund the introductory discounts being offered to new customers? The answer is clearly yes – everyone knows that dual pricing in insurance has been around for years. The key question is whether it has gone beyond acceptable limits.
Mary Starks, the FCA’s Chief Economist and Director of Competition, was clear on this during a Q&A session at a Social Markets Foundation (SMF) seminar earlier in the year. While dual pricing was acceptable in principle, she warned that if taken to “…too big an extreme, you do get outcomes that look iniquitous.” The FCA would then feel obliged to act.
The Evidence
The FCA has clearly been concerned about the level of dual pricing in insurance for a couple of years now. Consider a presentation by a senior FCA executive back in June 2017. Most people would judge the build up of a 70% price differential over 5 years between new and existing customers as ‘too big an extreme’. And I would include many insurance people in that as well. It simply fails to pass any reasonable test of fairness.
That 70% price differential raises an obvious question as to how customers respond (or more importantly, not respond) in such circumstances. Again, the evidence of the FCA’s approach has been clear for a while. Their behavioural science unit has been expanding its team of data scientists for a couple of years now, in order to undertake ‘large scale randomised control trials’. That slide above is just the tip of a much larger evidential iceberg that underwriters are about to be confronted with.
What insurance firms should be on the lookout for
What this points to is a regulator equipped with data and analysis of considerable depth and to a high standard. They will know an awful lot about how customers think and what makes them act (or more pertinently, not act). And all of this will be cross matched with those regulatory commitments to ethical issues like fairness and integrity.
Wrapped around all that evidence will be questions, and these will fall into two categories. The first category of questions will be around ‘why are you doing what you are doing’. And the second category will be more mechanistic – what data are you basing these pricing decisions on, and what parameters are influencing the analytics behind them?
Both categories of questions will be raised within one overall context, and one which individuals with pricing responsibilities should think seriously about. It is of course the Senior Managers and Certification Regime (SMCR). When it goes live towards the end of 2018, the SMCR will draw clear lines of responsibility from outcomes, through the decisions that drove those outcomes, to the individuals who made those decisions.
So the senior executives who have told me that they have to do dual pricing because ‘everyone else is doing it’ will be expected to justify the outcomes so generated against the context of regulatory principles such as acting with integrity. Given that integrity means doing the right thing even when it’s not in your interests to do so, that could turn out to be a tough call. Those five words ‘everyone else is doing it’ are not called the five most dangerous words in business for nothing.
If there is a previous FCA piece of work that firms should take as a template for how this scrutiny of dual pricing in insurance might be conducted, it is the regulator’s 2014 market study of the short term credit market. There are several similarities in both the problem and the approach.
How insurers might prepare for all this?
The starting point for how insurers should prepare for this review is scope. The review will start with dual pricing but encompass price optimisation as well. It will go into the data and analytics that are driving those pricing decisions. It will address issues like fairness, vulnerability, equality and privacy. Insurers need to understand this links in order to prepare a response that the regulator can judge informed and reasonable. You only have to read previous reviews to see how the FCA views insurers who just don’t see the problem being put to them.
Insurers also need to be clear on the principles that underpin their pricing strategy. Again, Mary Starks was clear on this at that SMF seminar. She talked about the FCA overseeing pricing practices on two levels: firstly, “do we understand the principles on which firms are pricing, and are we OK with those principles?”. And the second involves “…keeping enough of an eye on the data so that something that starts out as an acceptable principle doesn’t get taken too far”. So insurers need to have clear principles and be able to demonstrate that they’re an active component of how the business is managed on an ongoing basis.
Insurers then need to have ready for the FCA a clear template for how those principles are being applied within all those pricing algorithms now in widespread use. What signals are those pricing algorithms looking for in amongst all the data that insurers have accumulated about us? Many US state insurance regulators have been clear about what they judge to be signals that are unethical. Insurers in the UK need to be absolutely clear on this too, and be able to show how judgements were made and how actions flowed from them.
Insurers should not underestimate the FCA on principles and data. They may have shown signs of uncertainty in the past, but for the last couple of years, they’ve been in active partnership with the UK’s foremost institute on the ethics of data and artificial intelligence. Some of the data scientists recruited into the FCA’s behavioural science unit have been senior enough to be given ‘visiting fellow’ status at that institute.
What I would also advise insurers to start collating is all the evidence they have for having tested their pricing systems in relation to any unfair or discriminatory outcomes. Here’s what Mary Starks had to say on this during the Q&A session at that SMF seminar, and note that she was talking specifically about general insurance:
“There’s quite an interesting debate around the principles that firms are using when it comes to big data and artificial intelligence and machine learning and so on, because the firms don’t necessarily know what principles their algorithms are using – it’s just throwing computing power at patterns and finding stuff, and so there’s some quite interesting AI ethics questions around how do we ensure that firms have enough control over their algorithms to make sure for example that they’re not racist, or colluding illegally, or doing any of the things that we would object to a person doing”.
This is a message being picked up by civil society groups, many of who attended that SMF seminar, and many of who are specifically interested in general insurance pricing. Unfortunately, I was the only insurance person there.
How significant will this FCA review be
How significant will this FCA review of pricing be for general insurance? I strongly suggest that insurers do not underestimate it. All that ‘diagnostic work’ and that tie in with the UK’s foremost institute on data and ethics point to a huge (and expensive) amount of preparation. They must have scoped the problem as big enough, and significant enough, to warrant all that investment.
And there will be lessons in the outcomes that it generates for other market segments to take on board too. I’m thinking particularly of the life and health insurance markets, where pricing decisions can attract far more ethically charged issues than household insurance. Those other markets should take note of the FCA’s preparation, its framework of principles and its expectations of how insurance should work.
