Many people working in motor insurance will have been surprised by Citizens Advice’s research into discriminatory pricing. It will feel like something they and their firm just wouldn’t do. Yet the evidence that something is happening is there. So how do insurers respond to their obvious questions?
There are many good people working in insurance, across underwriting, claims, marketing and counter fraud. I’ve worked with many of them, particularly when I was head of insurance at Motability, Europe’s biggest vehicle fleet. These people will be wondering what has been happening to produce the results that the Citizens Advice (CA) research found. And they’ll be wondering what their firm, what their sector, will be doing in response.
Four questions will have been asked over the last week or two since the report came out. Something along the lines of… Is this actually the case? How can this be? What’s brought this about? What can we do about it? In this analysis piece, I’ll be providing some answers to these four questions, for insurers to make use of in their dialogue with employees.
Think about it carefully
What I think is pretty vital is that insurers should not adopt what Daniel Kahneman describes in his book ‘Thinking Fast and Slow’ as system 1 thinking. In other words, tackle the questions that their motor teams will be raising with a response that is fast, instinctive and emotional. Something along the lines of ‘how dare they’, ‘we would not do that’ and ‘they don’t know what they’re talking about’.
Instead, insurers should adopt system 2 thinking – slower, more deliberative and more logical. Sure, it will take longer, but it will also resonate more convincingly and produce a long lasting impact. The implications of the ethnicity penalty are immense. Insurers need to respond to them with care.
One option that might tempt some insurers is to simply ignore the issue and get on with business as usual. That would be unwise, for the regulator has already been told off by Parliamentarians for not engaging proactively enough with the sector on discriminatory pricing. And that was nearly three years ago. Doing nothing doesn’t seem to be a realistic option.
Sure, if an insurer’s monitoring for discriminatory outcomes in its pricing is long running and it has evidence to show that the issue has been addressed, then they can communicate all this to key stakeholders and indeed get on with business as usual. I’m not sure that many insurers are in that position though.
So, let’s examine those four questions I outlined earlier, being ones that employees will no doubt be raising now in some form or another.
Question 1 - Is this Actually the Case?
For many motor insurance people, the CA research will be uncomfortable reading, out of kilter with how they see themselves and the work they do. So a natural reaction will be for them to ask something along the lines of ‘are they right?’
As I said in this earlier analysis, I found the CA research to be pretty robust. So at the moment, the answer to that ‘are they right?’ question definitely sits somewhere between yes and no. As with any piece of research, its veracity lies in its reproducibility.
This means that similar research should come up with something along the same lines. So the question then evolves into ‘do we have evidence that contradicts or supports the CA results?’ Now, discrimination has been widely recognised as a data ethics risk for some years now, so that sort of evidence could well be found in the risk management programme that’s wrapped around your digital strategy.
Have your risk management people been monitoring for discriminatory outcomes in its pricing? If they have, then that’s the starting point for the firm’s response to this key question that employees will have.
If such work hasn’t been done, then the next place to look will be in the firm’s diversity and inclusion programme. What commitments, plans and reporting does that programme have in respect of non-discrimination and customers? There will be lots about employees, but what about customers?
Should neither of these sources prove to have relevant evidence one way or another, then clearly the conversation needs to move on to ‘why haven’t we done this’ and ‘how can we put it together now’. It won’t only be employees interested in this. Investors and board members will want to see the evidence that’s backing up any reassurances being given out.
Question 2 - How can this be?
This next question is a little more complicated. Readers of the CA report will notice that they do not set out any specific reasons for what they think is causing the ethnicity penalty. So how can firms think this one through so as to arrive at a convincing answer for employees?
Research published earlier this year found that organisations were aware of the ethical issues associated with data and analytics, but that in general, this has not yet filtered through to corporate governance frameworks. So it has been on people’s radars but out at the edges, not yet on to-do-lists.
The one inroad that data ethics issues have made into corporate governance has been around privacy, with the GDPR being an obvious factor in this. And we can see this amongst those UK insurers who are already building resource on data ethics. I can't think of one that does not position data ethics within its group privacy function.
Yet as the CA report reminds us, there’s more to the ethics of data and algorithms than just privacy. Indeed, if firms think along 80/20 lines, privacy should definitely be the 20 part, definitely not the 80 part.
The Influence of Legal
So then the next iteration of thinking about this ‘how can this be’ question points, in my opinion, to the influence of legal people. Their thinking tends to be orientated around what the law requires and privacy was given a lot of legal / corporate attention as a result of the GDPR.
Hold on a minute though, some of you will be thinking. How about equalities legislation? That’s been around for longer than the GDPR and DPA. What hasn’t that been given attention?
Well, clearly it has, in relation to a firm’s employees. Yet clearly, it looks like it hasn’t, in relation to a firm’s customers. That’s the blind side the sector seems to have had.
I recall encountering it some years ago. An insurer had announced that it would no longer be writing motor business in some of the postcodes of a particular city. And that city, and one of those postcodes, happened to be where I lived after I got my first job in insurance. Sure, the area was a bit rough around the edges, but on the whole, it was a nice, safe, diverse part of that city and I have great memories of those times. Yet the insurer didn’t see it that way. And I thought, what would I do for motor insurance if I still lived there. What would people still living there now do? It didn’t seem fair.
So I believe another factor behind this ‘how can this be’ question has been a mix of reduced competition due to patterns of ‘no-writing decisions’, and ‘go-away pricing’ for those still in the market.
There’s more we can say along these lines, but for the moment, let’s wrap up this question with some actions that employees can expect their firm to be taking. Here’s three…
- do a rigorous data ethics risks assessment so that you can get a better view of the overall data ethics landscape.
- engage with consumer groups to broaden the voices you’re listening to, so that how you interpret that landscape has just as much common sense in it as insurance sense.
- have a plan to make use of what you learn from those conversations, otherwise they won’t engage with you again. In short, move your horizons beyond privacy.
Question 3 - What’s brought this about?
In my opinion, there are two fundamental reasons for discriminatory outcomes having come about.
Firstly, a lack of understanding about the implications of ever granular underwriting. It doesn’t take rocket science to realise that the rapid and ever increasing granularity of data used in underwriting was going to exposure the sector to accusations of discriminatory practices.
The caveat here of course is ‘unless we operate a strict set of compliance controls in relation to equalities legislation’. And in my opinion, that does not seem to have been happening (more here). So as granularity increased, and micro-proxies proliferated, discriminatory outcomes were largely left to accumulate.
The second reason links with the obvious response to that last point – why were they largely left to accumulate? Why didn’t people realise what was happening? This brings us to the culture of firms and the micro-cultures within teams. I believe too many pricing teams suffer from conformity (more here). The culture tends to be too inwards looking, too easily swayed by calls sometimes heard from within the sector about ‘having to do whatever it takes’.
And part of why this has come about is down to pressure. Pressure on performance, on competitiveness, to be top of the fold, to push down that COR, to do something about fraud. And so you hear more and more rationalisations – the excuses people use to make a poor decision look better than it really was. At every training session I do on ethical decision making, I ask attendees if they’ve ever heard the phrase ‘everyone else is doing it’ at work. And every time, most people have.
The Most Dangerous Words in Business
Yet remember that ‘everyone else is doing it’ has been described by one of insurance’s most famous CEOs as the five most dangerous words in business (more here). Not far behind them in the ‘dangerous words to use’ stakes is ‘the company needed this from us’. And the appeal to loyalty upon which this phrase is based seems very isolated from the ‘appeal to the firm’s reputation’ that should have counter balanced it.
Let’s again wrap up this question with some actions that employees can expect their firm to be taking. Here’s three…
- deliver some training in ethical decision making and ethical dilemmas. Your people need to know not just what the right thing to do is, but how to do it.
- make sure your people know how to handle a whistleblower properly. I expect to see an upturn in serious concerns being reported and it’s important that they’re handled correctly.
- do more exit interviews. There’s a risk that some people will simply decide that this is not the job for them anymore and leave. Finding out their reasoning for doing so is an important learning loop.
Question 4 - What can we do about it?
Let’s bring Daniel Kahneman’s work back in. As I outlined earlier, insurers should avoid system 1 thinking that is fast, instinctive and emotional. Responses for example along the lines of ‘how dare they’, ‘we would not do that’ and ‘they don’t know what they’re talking about’.
Instead, they should approach this situation with system 2 thinking – slower, more deliberative and more logical. Responses for example along the lines of ‘how good is their research’, ‘could there be other factors at play here’ and ‘what do we know already about our pricing’.
Insurers need to give themselves time and space to see the wider picture, in terms of how this position arose, where exactly they sit in it and what can be done about it. They need to listen to people, and to talk to people.
I read the CA report as allowing the sector a little time to absorb the situation. After all, all of the report’s recommendations are aimed at the regulator. That said, time is not unlimited. The regulator could well be knocking on insurers’ HQ doors soon. After all, the Treasury Committee told them off back in 2019 for only accepting insurer reassurances about the compliance of their pricing with equalities legislation. Parliamentarians wanted the regulator to check the numbers upon which those reassurances were based.
Insurers should also communicate their plans for addressing the issue of discriminatory pricing with their board and senior management function holders. After all, the latter’s exposure to individual accountability under SMCR feels like it's just risen a notch or two.
Do the same with your pricing teams, and with any other teams whose work influences pricing. And make sure that dialogue is open and fair (more here). You want to learn through listening as well as show how you plan to respond.
Listening to Partners
The same needs to be done with your partners – those who you do business with upstream (suppliers, data brokers etc) and those who you do business with downstream (for example, scheme customers). The former may well be big influencers of the data and analytics you use in pricing and clearly, they need to have been providing you with services that complied with equalities legislation.
Reach out to any scheme customer who might be particularly interested in the CA research. Some charities place a lot of scheme business in the insurance market. They will be waiting to hear a) whether this has been happening to them and b) how you stand both in words and deeds on it. Let your account executives know about this – they will want to be managing those client relationship with care and understanding.
Let’s wrap up this last question with some actions that employees can expect their firm to be taking. Here’s four…
- run special datasets through your pricing system to establish where it sits in relation to the CA findings
- engage with data brokers and software houses to understand how they’ve been complying with equalities legislation
- listen to scheme customers stakeholders and communicate your perspective
- share your assessments and action plans with pricing teams, boards and investors.
Back in 2018, research by law firm DWF found that 28% of insurance executives thought that by 2022, the greatest risk they expected to face would be personal regulatory sanction due to their firm’s compliance failings. It feels like the Citizens Advice report on discriminatory pricing in UK motor insurance has brought that risk more to the fore.
Citizens Advice look to be in this campaign for the long haul. Insurers and insurance executives would do well not to underestimate this.