The Bank of England published a research report last month on ‘The effects of subsidised flood insurance on real estate markets’. Given the influence of the Bank on policymaking and regulation, this is the sort of research that gets widely read and influences decisions. It therefore has ‘trickle down’ significance for underwriters and public policy chiefs in UK insurers.
The big picture against which the report is set is about ‘transition risk and wealth redistribution in response to public interventions addressing climate change’. The research is however much more focussed – how did Flood Re affect property prices and transactions volumes in areas of high flood risk?
It is also a report full of assumptions that leave it open to pretty direct challenge. Those I will look at in a minute, but firstly, let’s look at what the researchers found in relation to prices and transactions.
“Our results suggest that Flood Re mitigates the negative effect of flood risk on property prices and transaction volumes.”
That sounds exactly what policymakers were hoping for: improvements in the value and liquidity of properties at flood risk in England. The effect was more nuanced however. The researchers found that Flood Re benefited those in wealthier areas the most. This was a distributional consequence of Flood Re that policymakers had not intended.
The Bounce Effect
This was an economic study, and I am no economist, but it strikes me, as someone on the Clapham omnibus, that the ‘bounce effect’ of Flood Re was always going to be in areas with wealthier populations. This is for the simple reason that higher property prices have more potential ‘bounce’ in them. Less well off areas generally have lower property prices and larger numbers of rented properties.
One could therefore just file the report as having interesting and relatively expected results. I want to explore it in a little more detail though, for as a ‘staff working paper’, this is work done by the Bank, not done for the Bank. It is revealing of Bank thinking on (and I would go so far as to say understanding of) insurance and the key trend that is defining its near future: personalisation. There are three aspects of the research I want to highlight: weak data, bounce back in 2039 and understanding key trends.
Weak Data Foundations
The report is very strong on property price data. It sourced individual transaction details directly from the Land Registry. This gives a good amount of confidence in its key findings about the effect of Flood Re on value and liquidity. Unfortunately, the report is very weak on insurance price data. This is one of the report’s key findings:
“Flood Re is estimated to reduce average annual insurance premium of flooded properties from around £650 to less than £325.”
In a footnote to that finding, the researchers acknowledge that “information about average house insurance premiums of flooded properties is limited”. And their estimation of a 50% reduction in the average annual insurance premium of flooded properties is based upon DEFRA data from 2013.
That date is significant, for it was when the ‘Statement of Principles’ (the pre-Flood Re agreement on flood underwriting) was extended for three years. Flood Re went live in 2016. During those three years from 2013 to 2016, it is my understanding (from various sources and cases) that household premiums for homes in areas deemed to be of high flood risk went through the roof. I mean premium increases from one renewal to another of five to ten times. I recall meeting with one couple (no claims for anything, let alone flood) whose renewal had gone from £350 to £5,000.
Such increases created crises of confidence in many affected communities, to the point that the Scottish Government commissioned research into the implications and its exposure.
Availability, not Affordability
Such pricing was entirely within the Statement of Principles. The statement was all about availability, not affordability. As a result, insurers instituted exponentially big increases in renewal premiums between 2013 and 2016. Relying therefore on 2013 premium data paints a very different picture to what people were actually experiencing.
The impact of this is that when Flood Re went live, average premiums for properties in areas deemed to be of high flood risk dropped considerably more than 50%. I would estimate the actual figure to more likely be in the 200% to 500% range. This testifies to the enormous impact of Flood Re, not just for flooded properties, but for any property in an area of high flood risk. The researchers see the latter as about 100 times more than the former. That difference matters, for the impact of Flood Re is circa 100 times wider than just those properties with a history of having been flooded.
Risk Reflective Pricing
And that difference really does matter when you align it with two developments. The first development is the end of Flood Re in 2039. That is the date by which flood underwriting is to have transitioned to ‘risk reflective pricing’. Or to think of it another way, it is the date on which the prospect of the reversing of those enormous premium reductions rears its very considerable head. And that prospect will, as the researchers found, be most tangible for those living in wealthier areas.
Sure, insurers will say, technically there is such a prospect, but at the same time, government investment in flood defences, and personal investment in flood resilience measures, will go a considerable way to counter balance that prospect. I’m not so sure that the market will think in that way as 2039 approaches, the reason being the second of those two developments.
That second development is personalisation, by which is meant the ever granular underwriting of individual risks. Flood Re is a case study par excellence for what can happen when personalisation is fully implemented. Between 2013 and 2016, the level of granularity in flood underwriting increased significantly, to the point that by the launch of Flood Re, it was being done on pretty much a property by property basis. It was only fair, went the narrative, that those living in areas of high flood risk should pay premiums reflective of that risk, even if they were, on average, one of the 99 out of a 100 whose property had never flooded.
The Prospect of Bounce Back
When Flood Re ends in 2039, will people living in areas of high flood risk experience premiums pretty much in line with Flood Re levels? Only if all the flood defences were in place, and only if insurers actually took account in their rating of flood resilience measures taken by the policyholder, and only if they differentiate between the, on average, 1 in a 100 and the 99 in a 100.
There we have a problem. I would expect most but not all flood defences to be in place. I would expect the capacity of flood events to have increased, thereby rendering some (probably older) flood defences only partially effective. I do not expect insurers to actually take account in their rating of flood resilient measures taken by the policyholder – insurers rarely do so unless wrapped up in a warranty type endorsement. And I do not expect insurers to really and truly differentiate between the 1 and the 99 in a 100.
The risk of premiums bouncing back to those experienced in 2013 to 2016 is considerable. Premiums to cover flood insurance will, I believe, soar again in 2039. It will feel like we’re back at square one. Time for a return of the Re?
Those are in a sense the practical concerns I have. Just as significant in the medium term are signs in the report that the Bank of England does not really understand what is happening in insurance. The researchers appear to be working with a mindset that a price that is not fully individualised is a price that is unfair. The title of the report talks about ‘subsidised flood insurance’. In the introduction, they refer to the premiums passed on to Flood Re from the fronting insurer as being ‘highly discounted’.
The pricing of the flood insurance provided by Flood Re is not subsidised, nor is it highly discounted. It is simply based upon a pricing methodology (pooling) that is different from that around which personalisation is being based. And that pooling methodology is one that was current in the market from roughly 1800 to 2010.
To say, in essence, that a pooled price is subsidised and personalised price is not (as the report’s language infers) is an interpretation that I don’t believe Bank of England economists can make, at least not without reference to it being an interpretation that is open to debate.
A Reminder of 2008?
OK, so some of you may think, am I making a mountain out of a molehill? I think not, for the simple reason that the Bank of England is a very influential body in financial and political circles. If it is producing reports that in a particular research niche make sense, but which in the wider landscape of how insurance is changing do not make sense, then I believe it is beholden upon the Bank to take steps to at least recognise, better redress, that imbalance.
It's worth noting that the Bank of England is not alone in this. The Financial Conduct Authority is just as susceptible to adopting a ‘personalisation good, pooling bad’ way of thinking. What both lack is an appreciation of the wider debate that is emerging around how exactly insurance can evolve within the varied opportunities and risks that its digital evolution is presenting. Without such an appreciation, there’s a big risk that they will fail to engage, understand and influence the big changes happening in the market at the moment. This may sound harsh, but it reminds me a bit of 2008.
Insurers need to both analyse the report's findings and position them within a wider, more nuanced and complicated debate about the future of insurance. Even insurers with little presence in the household insurance market need to do so, for flood insurance is a bellwether for more significant developments happening around life and health markets.
This means public policy people, digital strategists and underwriting strategists not just driving change, but also understanding where that change is taking them, understanding where their firm will end up in the wider landscape of a digital insurance future.