- a mystery shopping exercise found that in each of the areas they tested that had a high proportion of Black and South Asian people in the population, customers were quoted at least £280 more for car insurance, compared to areas where the population is largely White.
- that same mystery shopping exercise did not find any statistically significant differences in prices paid by customers with different names.
- an analysis of over 18,000 people who came to Citizens Advice for support with debt in 2021 found that people of colour report spending on average £250 more than White people for car insurance.
This is what they’re calling the ‘ethnicity penalty’ and it's being positioned as a significant campaign, starting with insurance.
Is the Research Robust?
I had a long, careful read of the research paper and to be honest, I found it pretty impressive. The structure of the research is balanced, involving both a broad base of individual data and more detailed investigation of a population sample. The former was collected and analysed by Citizens Advice itself and the latter by an independent consultancy.
The analysis of the findings involved a range of control factors that I found convincing. For example, not only crime rate, but also deprivation, traffic accidents and traffic density were controlled for. Account was also taken of the influence of counter fraud measures on the quotes obtained through mystery shopping.
And I also found the section ‘Why is this happening?’ to be realistic, recognising how these different quote prices might have arisen and the extent to which insurers can control for them.
So, overall, I find it a good piece of research. The findings need therefore to be taken seriously.
What is being asked for?
Citizens Advice want explanations, from insurers on their pricing practices, and from the regulator on how they’re overseeing this. Here are the five steps that they want the FCA to take:
1. Publish a public statement setting their expectations for how firms should demonstrate that their pricing practices comply with the Equality Act (2010) and their obligations under fair pricing regulations, and what action will be taken against firms who fail to meet these standards;
2. Require firms to audit and monitor pricing outcomes to identify any racial disparities, and to cross-check permitted data for correlations with protected characteristics, and report these findings to the FCA;
3. Conduct work to measure any correlations between profit margins and the racial composition of geographic areas that could result from pricing algorithms;
4. Take enforcement action against firms found to be in breach of their obligations, or failing to explain why their pricing models have delivered differential outcomes effectively;
5. Assess and build capability for effective oversight and monitoring of algorithmic decision making, to future-proof their regulatory approach as the prevalence of big data and machine learning lead to ever more personalised pricing.
We know that the digital complexity of insurance pricing is not going to go away, and Citizens Advice recognise that. What they do want though is for that complexity to be accompanied by an equal measure of transparency and explanation.
Are These Findings a Surprise?
There was nothing surprising in the Citizens Advice report. Since 2015, I’ve been saying that insurers needed to pay more attention to discrimination in their underwriting, claims, marketing and counter fraud operations. Here are some of the main pieces I’ve written:
- Back in 2015, I conducted a survey of corporate equality policies, looking for how insurers referenced equality in terms of their products and services for consumers. The results were very disappointing (more here).
- In 2017, I wrote that discrimination was one of the seven main risks associated with insurance pricing (more here).
- In early 2020, I said in this article that ‘discrimination was an enormous ethical challenge for insurance’ and set out steps that six functions needed to take.
- In late 2020, I highlighted the signalling on discrimination in the FCA’s final pricing report and examined the caveats to the exemptions for insurance within equalities legislation (more here)
- And in the ‘big challenges’ articles I write at the start of each year, discrimination was highlighted in 2020, 2021 and 2022.
I reiterate: some form of challenge to the sector on discrimination has been obvious for several years. Even more obvious than the super-complaint was.
OK, I’ll stop blowing my own trumpet and move on to analyse the implications of Citizens Advice’s challenge for the sector and its regulator.
An Approach Repeated
The approach of Citizens Advice to the ethnicity penalty is remarkably similar to that they took with the loyalty penalty. They draw on the huge amounts of micro-data collected with their network of advisors and analysed it for issues. They then investigated the issues they found through more detailed and focussed research.
Being evidence based, this gives their conclusions considerable weight. It’s hard for the regulator or insurers to argue against them. So in effect, like in tennis, they’ve served a strong ball that the other side will find difficult to return. First game therefore seems to have gone to Citizens Advice. I am certain that, as with the loyalty penalty, Citizens Advice have a game plan ready for the next year or so, with a lot of political lobbying and media resources prepared.
Reading between the lines of their five recommendations, I get the feeling that they’re pretty annoyed that the FCA have not done this work themselves. After all, it was the FCA’s then CEO, Christopher Woolard, who told the Treasury Committee back in 2019 during a review of discrimination in insurance pricing, that the regulator had “the resources and expertise to pick inside those insurance models”.
Clearly, the FCA have not followed up that bold assertion, despite having assembled a huge database of motor and house pricing as part of their review of that pricing super-complaint. This means they’ll have to deal with some harsh questions at their next sit down with the Treasury Committee.
It’s about the Regulator
So why haven’t the FCA done more to-date on discrimination, when it was so obviously in their interests to do so? Well, they told the Treasury Committee that they didn’t want to take responsibility for insurers’ compliance with equalities legislation, preferring that it remain with the Equality and Human Rights Commission. So what’s behind that?
Well, it’s almost certainly down to a blind spot that the FCA has in relation to anything that smacks of social justice. It took a super-complaint to make them see fairness in pricing as anything other than just something that the market did. So while the FCA and the EHRC signed a Memorandum of Understanding in early 2021, I expect that was for the FCA to support the EHRC with all that ‘resources and expertise’. Either which way, nothing has come out of it, so far.
This doesn’t cut much ice with Citizens Advice. They state quite bluntly:
“The FCA has a responsibility to ensure that firms treat all customers fairly. Any evidence of an ethnicity penalty should be a cause for concern, and should be investigated further. “
And they follow that up with a key point:
“While the FCA does not have enforcement powers under the Equality Act, they have previously stated that firms would not be meeting their obligations, including the FCA’s Principles for Business, if they did not comply with the Equality Act.”
Citizens Advice then drive this home in their recommendations. All five of them are written in relation to the FCA, not insurers. They read very much like a ‘please do your job properly’ statement.
You may well ask why Citizens Advice didn't orientate their recommendations towards the Equalities and Human Rights Commission. I suspect this was down to the FCA having previously investigated pricing and having given reassurances to Parliament that they had the resources to investigate discrimination.
The Issues for Insurers
Citizens Advice address the $64k question – what is driving these outcomes – pretty much head on. They recognise that decisions about insurance pricing are made in the context of widespread systemic racism:
“Systemic racism in areas like education, employment and housing has created structural inequalities for people of colour, which are reflected in many of the rating factors used by insurers to calculate risk. While individual rating factors may be considered legitimate, taken together they could contribute to the poorer outcomes people of colour experience.”
They go on…
“The insurance market operates in a way that incentivises consumers to minimise their personal level of risk, but when these risk factors reflect structural racism - which is not within consumers’ control - the system becomes unfair. Nor are these structural factors the fault of insurance companies, and it would clearly be challenging for the industry to mitigate all impacts of structural racism.”
The Critical Point
And they then deliver this critical point…
“Structural inequality embedded in risk rating factors may well play a role in determining insurance prices for people of colour. However, the factors we might expect to drive differential outcomes in the market cannot account for the findings of our research. We found that quotes in lower crime postcodes with large Black or South Asian populations are higher than quotes in higher crime postcodes with a largely White population. This suggests that the ethnicity penalty is not solely driven by inequalities caused by systemic racism, which are largely outside insurance firms’ control. Rather, insurance pricing mechanisms could be perpetuating or exacerbating discrimination in the market.”
Some of you will have worked with me on data ethics and remember that at the heart of the data ethics framework used are three components of data ethics risk: data, algorithms and practices. The point that that paragraph makes is that insurers may be in no different a position with regard to the structural racism inherent in all the data floating around society, but they are responsibility for what their algorithms and management practices then do with that data.
Note however that I say “…insurers may be in no different a position…”. I don’t say ‘are in no different a position’. That’s because Citizens Advice seem to be giving insurers the benefit of the doubt on data, for now.
Three Lines of Defence
In recent years, the things that have been shared with me in relation to fairness and discrimination in underwriting, claims and counter-fraud have led me to conclude that the three lines of defence suffer from fundamental weaknesses (more here). Why this is so can be summed up in two ways.
Firstly, there are inherent conflicts of interest at play in those three lines. And secondly, they suffer from ‘cultural capture’, which causes all sorts of things to be ‘just not seen’ (see these toolkits). So I fear that insurers may struggle to respond effectively to the challenge that Citizens Advice has laid down to them through those five recommendations. In other words, some insurers will struggle to see ‘the problem’ because the organisational structures they have relied upon to-date have not seen it.
To address the problem, insurers need to reform their three lines of defence as well.
The Long Haul
I got a strong sense from reading the report that Citizens Advice see this research as the opening stage of what might turn out to be a long haul campaign. For example, on p10, “Our research was exploratory...”. I wonder what they have up their sleeves for the next stage.
They’ve raised three reasons that they think are driving discrimination in pricing –
- complexity of insurance pricing
- the opaque nature of algorithmic pricing
- the broader context of systemic racism in the UK
What they’re doing is handing over the running of this, for the moment, to the regulator and standing back to see what happens. This is because Citizens Advice has an awful lot of other things on its plate at the moment.
They’ll now engage in two things. Firstly, some political lobbying, in part to make their case and in part to put pressure on the FCA. And secondly, they’ll listen and watch for who does what when.
The problem for the sector is that it doesn’t really have anything like the micro data to shape into an alternative explanation for what is happening. The ABI will make noises along the lines of ‘we are good people who would not do that’, but then struggle to provide the evidence that ‘we do not do that’.
Evidence and Impacts
On several occasions in recent years, I’ve made it clear to insurers that they need to build their evidence base on how their underwriting, claims and counter-fraud systems are handling discrimination. If they believe, as many have told me, that they ‘would not do that’, then they need to show the evidence that confirms that they are indeed not doing that.
Citizens Advice make this point:
“Monitoring proxy data to better understand outcomes for different consumer groups, as we have done in our research, could allow firms to prevent discrimination without needing to expand their collection of data on protected characteristics. However, there is little evidence that this type of outcomes monitoring is routinely carried out by insurers, and it is not a regulatory requirement.”
Such monitoring is not just for audiences like regulators and consumer groups. Boards of directors will now be just as interested to see that confirmatory evidence. And investors will want to know that it has been put together and scrutinised, given the impacts that could result from any bad news.
What might those impacts be? Reputation is a pretty obvious one. Not many consumer facing brands fare well after being found to have been discriminatory. Fines and compensation could be expensive, as could the removal of the sector’s exemptions under equalities legislation.
I am particularly minded however of this news article I wrote a few days ago, about the use of ‘algorithm destruction’ by a US regulator. Just think of the impact that a regulatory order to destroy a discriminatory algorithm would have on a business. It could shut it down. Sure, replacement algorithms could be bought in, but remember that they would only be allowed to become operational after being confirmed as non-discriminatory. So we're talking months, not weeks or days.
What now for the Regulator?
The report is not a super-complaint, so the FCA is under no obligation to react immediately. However, the seriousness of the research findings plus the bold assertions made to Parliament by their CEO back in 2019 create an expectation that they will need to respond in some way, in days rather than weeks.
How long they take to respond could be influenced by how substantial a body of work they can bring forward now. If it's thin, they'll take longer to build it up. If it is substantial, then we could see something either this week or next.
Either which way, looking forward, they will be under enormous pressure to show that they understand the problem and have a plan to address it. Sure, they might try to bat the issue over to the EHRC to handle, as in 'they handle equalities, not us'. That would not be a wise move, for the report talks directly to the FCA. Standing aside in such circumstances would create a PR disaster.
The most likely response will be a joint EHRC / FCA team built up rapidly and put to work. Nothing in the Citizens Advice report is beyond either organisation to replicate and expand upon. I expect the EHRC will shape their response in line with their investigation protocols.
What Insurers Should Do Now
Writing this within hours of the Citizens Advice report coming out means that this list of suggested actions is summary. I will undoubtedly engage with some of you in more detail, but for the moment, these are the actions that I think every UK personal lines insurer needs to take.
- Read the report, carefully, and with an open mind.
- Review your firm’s policies and procedures in relation to discriminatory outcomes for consumers from underwriting, claims, counter-fraud and marketing. How clear and detailed are they?
- Check on what your risk and compliance people have already been doing around discriminatory outcomes for consumers.
- Reach out to any consumer group with whom you have an existing relationship and ask for their views on the Citizens Advice report. In other words, get some second opinions.
- Challenge yourself on just how much an impact what you have already been doing has had. Now is the time for some honest ‘weighing ups’.
- Instigate conversations within your pricing teams, to help them understand what the Citizens Advice report is saying and to draw on where they think risks and opportunities lie.
- Build a discriminatory prevention programme with dedicated resource, reporting into a non-executive director. This really is that serious!
- Have your regulatory experts draw on their relations with the FCA and EHRC to get a feel for what they think you should be doing next. However, remember that they are probably also on the back foot at the moment.
What I would strongly recommend is not done, is the issue of any form of denial, along the lines of 'we would not do anything like that'. Sure, insurance is full of good people and discrimination is something they would not countenance. However, what counts here and now is the evidence each insurer has to confirm that they have indeed not been 'doing anything like that'.
This is a crucial moment for UK personal lines insurance. Citizens Advice have challenged the sector like it has never been challenged in a long time. The implications of the ethnicity penalty are much more serious than those of the loyalty penalty. Insurers need to be honest with themselves, challenge themselves and work to resolve the issues raised.