JBA's Managing Director in the UK, Jane Toothill, and Managing Director in the US, Matt Reid, recently joined InsTech London for the latest episode in its podcast series. Hosted by InsTech London's Matthew Grant, we explore the lessons we've learnt in our 10 years of modelling flood in the UK, and how these can be applied to the new challenge of creating the highest resolution flood data available in the US.
Listen to the full podcast below, or read the Q&A.
Jane: JBA Group started as an engineering and environmental consultancy specialising in flood risk management in 1995 and now employs over 500 people worldwide. I joined JBA in 2008 with the intent of setting up its business in reinsurance and catastrophe modelling.
JBA Risk Management was set up in 2011 with 18 people, and now employs over 100 people across six different countries. We've retained our original focus on flooding and we provide a range of products and services to companies throughout the financial and property sector, including insurers, reinsurers, governments and banks.
Jane: Yes, we do. JBA Consulting remains an engineering consultancy focused on everything to do with water. Its expertise allows us to complement the data and models we offer to clients with the full range of flood-related services, from desktop work such as flood models right through to site surveys, assessments, and even the design and deployment of flood mitigation works.
This means that a company with a portfolio of multiple locations that is seeking insurance might first assess its exposure across the whole portfolio using our maps and models. If the level of risk at any given location is particularly high, they can work with JBA Consulting engineers to obtain a more detailed site flood risk assessment and potentially go on to work with the Group on designing and implementing the measures needed to mitigate the flood risk.
Jane: We started our flood mapping work in the UK and Republic of Ireland. Once we’d mapped those countries, the challenge of generating a global set of flood maps was too much to resist and now for several years we’ve provided 30m flood mapping for every country worldwide so long as it’s not covered in ice.
In parallel with the mapping work, we started developing a global flood event set. Once we had those two datasets, releasing a global probabilistic flood model was the obvious next step for us. We launched that in 2019 and it caused us to completely rethink how catastrophe models are created. We had to sit down with a blank sheet of paper to work out the best approach for designing a catastrophe model for flood at that scale.
The result of that is our modelling technology that builds a flood model on the fly at runtime, selecting the flood frequency and severity data required for the analysis of that specific portfolio at a resolution that's most appropriate for the data.
It allows us to maximise analysis efficiency without compromising the accuracy of results and give users more control. For example, the user might choose to adjust the relationship between flood depth and damage at a property to adjust assumptions about the standard of protection offered by flood defences. They can also specify the hours clause to be applied at analysis time.
Jane: The financial sector is waking up to the problems that climate change will cause, particularly with flood where there's a body of evidence showing links between a warming climate and increased severity and frequency of flooding.
We started in the UK with the release of a climate conditioned catastrophe model, which has been available for a couple of years now. We followed that more recently with the release of our Climate Change Analytics data, which enables clients to look at the impact of different concentration pathways and different time horizons on the risk at a single location.
Jane: We took our current UK Flood Model and adjusted the frequency and severity of events for a chosen climate scenario and future point in time, which was 2040.
Having received good feedback on the steps we have taken with that model, we're now looking at how to extend the approach to enable clients to pick from a range of climate scenarios and time horizons for which they would like to consider climate change.
Jane: Our two main products, the flood maps and probabilistic models are used in quite different ways. The maps are primarily used for underwriting and pricing. They provide the water depth associated with different frequencies of flooding and derive flood risk scores and annual damage ratio data which summarise the risk at a location into a single number that can be easily translated into underwriting rules. That enables companies to apply different terms to policies in different locations according to the level of flood risk.
The catastrophe models are used to calculate the potential cost of flood for a portfolio of risks. The output is a loss exceedance curve which gives the expected cost of flooding at different return periods at either net or gross of insurance and reinsurance terms. That information can be used for risk management, enabling a company to manage its portfolio so that its overall exposure to flood doesn't reach a level that it finds unacceptable, or indeed that the regulator finds unacceptable. It can also be used to help calculate the technical cost of reinsurance purchase for flood risk.
Clients are using our UK Flood Model to compare the results to those from the future climate model. That gives them guidance on how they could expect to see their overall risk change in the coming years.
Jane: Flooding is quite complex as a peril and can result from several different causes, but the three main types are coastal, river and surface water flooding.
Coastal flooding is caused by high sea levels, driven by a combination of wind and low pressure. We have to take the state of the tide into account when we model coastal flood, because a large surge event may have no impact at low tide but considerable impact at high tide.
River or fluvial flooding is the result of water levels in rivers exceeding the capacity of the river channel or its defences, causing inundation of surrounding areas. Surface water or pluvial flooding is caused by excessive rainfall that lands on the ground and indirectly causes flooding.
There are other sources, for example, the failure of reservoirs or canals or the rise of groundwater. A tsunami would be another kind, in that case, caused by earthquakes, but the three main categories that cause the majority of flooding are coastal, river and surface water.
Matt: Personally, I believe the predominant flood peril in the United States is actually the “protection gap”. When we look at data from hurricanes like Harvey and Sandy, one of the things that stands out is that many of the properties that were impacted had no flood cover whatsoever.
That is bad news for the property owners, but it also carries a degree of reputational risk for insurers. Property owners expect their insurer to offer appropriate cover and when they fail to do so, it reflects poorly on the insurer. We've seen legal action in the aftermath of Hurricane Harvey that relates exactly to that situation. Homeowners are unhappy that they weren't advised by their agent or insurer to take any flood cover, but then found themselves under several feet of water.
On the predominant types of flooding, tropical cyclone activity on the eastern seaboard and in the Gulf States drives the majority of the economic losses. That’s what happened after Hurricane Sandy, with significant storm surge, and fluvial flooding was predominantly what we saw with Hurricane Harvey as the result of a massive amount of rainfall being dumped on the landmass.
Matt: The great Mississippi flood of 1927 unfolded over months and caused damage that would be estimated in the range of $1 billion today. That’s noteworthy because it speaks to the range of floods that we see in the US. It was the result of heavy rainfall across a massive catchment area and it took months for the water to move downstream to the point where the Mississippi River couldn't contain it.
Today we have flood management measures and mitigation measures to prevent events like that, but the risk is still there and we'd be foolish to forget it. Flood is the most pervasive of all the natural perils and occurs in every state. It's not just the large events that make national or international news that drive losses for insurance.
Matt: There are two aspects. The regulated market, which exists under the National Flood Insurance Programme (NFIP), and a private market that is not subject to the same degree of regulation.
The NFIP classifies certain properties as being in or out of certain flood zones, according to its flood insurance rate maps. A community that elects to join the NFIP has to undertake certain floodplain management activities. In return for engaging in that programme, there are fixed insurance rates set for flood covers.
We could argue the pros and cons of such a heavily regulated government programme. It is heavily in debt and the limits on the cover that the NFIP provides are often not adequate. This is why we're beginning to see the emergence of a private flood insurance market, although the penetration is still very low.
Matt: We've leveraged our experience in modelling flood at high resolution and scale. All of the lessons we’ve learned about “how” to model flood in the UK, and globally, carry over. Some of the inputs may change, but the fundamental processes and methodologies all hold.
When our maps are being used as extensively as they are in the UK, they get a lot of scrutiny. That feedback has helped us refine and tune our methods and approaches, and now we've brought all of that experience to bear in developing our US flood maps.
The second dimension is our experience in helping clients integrate flood data into their systems, workflows and decision-making processes. It's not just about having really good data. It's about knowing how to use that data and how to deploy it appropriately.
Matt: We offer US clients pretty much everything that Jane described. We have hazard data at a five-metre resolution which is translated into flood scores to make it consumable for underwriting. All of our hazard maps come with a defended view and an undefended view, which means users can assess risk based on flood defences that are in place, or on a worst-case scenario where all of the defences fail. Some flood defences in the US are not as well maintained as they are elsewhere in the world, so having that flexibility is important.
We do also offer Pricing Data for the US, which offers the ability to query an individual location for an average annual loss or a technical price for flood. We have the capability now to provide probabilistic flood models for the US as well and potentially, given the flexibility of JBA's modelling technology, for individual states.
Matt: From the from the US point of view, the one thing we've not touched on is the fact that our hazard data being at five metre resolution is, we believe, the highest resolution continental scale flood data available. With a peril like flood, that resolution really does matter.
Also, while storm surge is a major factor in flooding, it is a wind-driven peril and we are ‘the flood people’. To do the best job we could of modelling storm surge, we partnered with a US company called Applied Research Associates (ARA). They have one of the four commercial hurricane models that are approved by the Florida Commission on Hurricane Loss Projection.
We use ARA’s data to understand the maximum sea levels driven by those wind events and we model the inland flood impact from that point on. We’ve combined ARA’s expert knowledge of wind and how it drives sea levels with our expert knowledge of modelling flood to get the best possible data.
If you're interested in learning more about our US flood data, get in touch with the team.
We're proud to be a partner of InsTech London - you can find all the latest podcasts via the website.