What can financial services tackling TCFD learn from insurance catastrophe modelling?
Finextra Research (FR) - We’re here because we were talking about the data aggregation and risk community that Oasis Hub run and a lot of what we've been talking about on sustainable finance is how do we get some of this fantastically granular alternative data from satellites etc into the world of risk management, ESG and so forth in the banking industry.
We started talking about what you've been doing in the insurance industry, so that's what we want to get into today. First of all tell me a little bit more about you. How did you get into this role, what's your background and can you describe Oasis Hub?
Tracy Irvine (TI) - I've been in environmental science for the last 30 years and I'd been working on project management of a European-wide project around climate climate science and climate services, and this led to the formation of our company Oasis HUB. We started as a result of a H2020 project that was a project that linked both the insurance sector through a Oasis Loss Modeling Framework who have a collaboration of 40 insurance companies and 20 universities who were sort of developing climate data climate and climate science. As a result of that the insurers were saying it would be great if we had a global hub around data and tools about what's available because there are thousands of different data sets globally being produced by universities, and being produced by companies and why not pull it all together so we know what's available.
Then on the other side the academics were saying because of climate change we're actually doing a huge amount of new science but we really want to get some of our data used and have bigger research impact of that data so that type of hub would also benefit us to bring data to the market more quickly rather than being sat back in their own databases, so that's how Oasis Hub came about. We were supported a lot by European Commission funding to develop the hub. We launched in June 2017 and since then we've been bringing thousands of datasets from across the world that anybody can use. Primarily we have certain markets that are looking to assess physical risks so the type of data that we hold is that of catastrophe, climate change and risk data. We're pulling those data sets together at many different resolutions so that anybody can find the data that they need.
FR - So you've come from that insurance background and now with Oasis Hub you're targeting people like the banking industry, risk and ESG etc. Is that right? That's the new market for you?
TI - Yes, we look very much at what the end users needs are. The insurance sector have been doing risk assessment for the last 20 to 25 years and have become quite advanced in the methodologies they use, in such things as developing full catastrophe models. They include not just the physical risks but also some of the vulnerability of communities they're looking at, and how good they are and then bringing in some of the financial modelling into that.
Whereas in other sectors, the banking sector is relatively new to this area, obviously the Task Force on Climate-related Financial Disclosures has really been a catalyst to getting financial institutions to think about their potential short term and long term risks on their capital allocation. The banking sector right now is quite a way behind the insurance section I think to certainly learn from the insurance sector in the way that they understand risk. But at the same time they almost have an opportunity to learn from the mistakes of the other sectors and do it better.
We will also provide data to banking downstream companies, such as property companies, commodity companies or energy companies all looking to understand their risks or their particular products and portfolios so the task force has essentially acted as a real mechanism to push climate risk assessment down into all levels of financial and societal management.
FR - Banking is a very broad church - retail, commercial, investment, capital etc and many variations within each of those. Quite a lot of the time when they're starting to tackle the TCFD requirements, they're finding that the data needs overwhelm them a bit. They've never had to deal with this before. On one level some people think they shouldn't go into this or they'll be getting a reputational risk if they do it incorrectly. Some are going perhaps a little too fast and get accused of greenwashing.
Did the insurance industry go through a similar journey? Was there this reluctance and hesitance or did they all go we're going for it all in one go?
TI - I wasn't there at the beginning but from what I've read there was a company who created the idea of looking at risk in terms of catastrophe modeling and then it's very much involved in science and what science tells us about risk. That was the feeling behind it that risk assessments need to be based on the fact and the best science at that time.
It's taken 20 to 25 years to make models more and more sophisticated, stronger and stronger in terms of how their uncertainty levels are. Nonetheless there still is uncertainty, but they continue to improve as the science improves. What the insurance sector have managed to do is track how good the sciences and track how that's developing and put that into their models and I think that is the approach that anyone should be taking. You can do the best on the best information you have now but it will keep improving in the future. Whatever services you invest in you need to be asking questions - are you regularly updating this data, are you in touch with organisations that are going to improve this system. Those are the questions I'd be asking.
Over the next ten years it's probably going to be a case of seeing rapid understanding of climate change and the impacts of that on societies both economically, physically and financially. We have to remember that there is not just one hazard you're looking at but one hazard can then turn into cascading hazards. This is a very complex area but there's no reason why you shouldn't start now. Through statistical processes you can factor in what's average risk, what's exceeding risk, what's exceeding the kind of norms and how often that exceedance is actually happening. That's what's going to guide you know where your climate change tracking is going.
FR - One of the debates in the banking industry is that the data isn't there and we haven't tracked it and if we now start to try to bring it in, the data I get from one supplier is doesn't score the same as the data from another supplier. It's a kind of ‘oh my goodness it's all so complicated world', so there's there's a reduction to a standard mindset going on so that at least you know they can benchmark apples versus apples.
The feeling is when you when I hear you talking that there's obviously such nuance in in climate models you know and in geospatial data, down to the the physical assets, whether it's the UK or whether it's you know London - different geographic spacing. How did the insurance industry, how do your first banking customers bring together these two spectrums? How would you encourage people to go ahead with this?
TI - The way I would probably do it is, first of all in broad terms, ask what are the global data sets saying, and what are they saying for my particular location. Realising that they are low resolution data and that they have a limited ability to understand your risk. Beyond that then you need to be thinking about what sectors do you have that could be particularly at risk and then perhaps look at how you kind of model those in much more detail so that you understand the risk in more detail. Although I understand that the financial sector use a lot of spreadsheet risk assessment I think you should probably go hand in hand with a geospatial platform so that you've got a verification tool of what your you know your financial systems are telling you. I think the human the human eye can do that quite effectively.
In terms of the variation of the data, the way it works in science is that is the scientists are comparing the datasets that are that are provided and if those datasets are massively different, which some will be then they start to ask the question 'what have we done wrong' on both sides. That's how the science will gradually improve until you get to a point where people have a higher certainty on the types of impacts that might happen, how to measure those impacts, where are those impacts going we still don't know that although we have the climate change scenarios these are just you know set at a level that scientists think are feasible but they could overshoot there, they could be less than that. So understanding your exceedance risk is the main thing and then work down from there.
FR - You're there as a aggregator in a community. You bring datasets in from many different sources. You're a one-stop shop for many people. Do you provide those geospatial tools that people can use for analytics or do they plug it into their own ones? How does it have the distribution of the data happen once you you've acquired it?
TI - We've got a range of different things. We've got the very data themselves, so if you've spent the time creating a system internally for your particular portfolio, you can take some of our data put it into your internal system.
What we're beginning to see and hope to grow into is a toolkit of different tools that can look at specific risks. We've got quite a few tools that look at flood risk, coastal flood risk, some tools that look at wildfire risk, at both larger scale but also local scale. If you were investing in some you know very high value you know property or industry or infrastructure you might want to look at some of the more localized tools that you can use so you get a specific view on that risk.
Thirdly we have the insurance development forums catalog of catastrophe tools. That's got two-thirds of the world's catastrophe models listed on it so if you're interested in understanding your full portfolios on a much larger regional basis, most of these tools have you know thousands of different events modelled into those models and thenthey will be able to be put through a calculation colonel such as the Loss Modeling Framework with oasis. They will be able to give you the physical risk but also the financial risk or the financial damage likely to be caused by that risk at different scenarios.
There's a lot of different tools that and I see Oasis Hub becoming more like a toolkit for specific needs. That's one of the things that we're trying to get people to understand is that you really need to understand what you need from risk assessment and then compare the market on what's the best approach for your particular type of risk assessment. I think there's too much at the moment trying to make the environment into this flat playing field. It would be great if it was like that but it but it simply isn't like that.
FR - What's the future? Are you using already things like Artificial Intelligence or is that something you're heading into? What's happening on the tech side in that area?
TI - Artificial Intelligence is certainly one of the big things being used by both the satellite industry and the drone industry, looking at damage assessments. All of these, AI, identifying risk, identifying what type of housing portfolios there are globally etc. There's loads of tools that you can look at and use for whatever you know purpose you want in terms of risk assessment. All of these things are coming through. The thing with Oasis hub is one of the things we kind of pride ourselves on being is a collaboration portal between the demand size and the producer size of this data, and trying to pull them together. We have projects where say we've pulled insurers together with scientists where both have learnt from each other in terms of how to assess how to assess risks for you know the best societal purposes.
We act as that kind of collaboration access portal. At the same time scientists are bringing in new products new tools that they can do. There's brand new ones coming on in the autumn that are looking at adaptation. You can look at your flood risk and you can look at the different scenarios and then you can look at how you might be able to adapt to that risk. There's really amazing tools coming onto the market. Similarly in forest fire, looking at multiple meteorological conditions that will influence the fire and you as the user can ask questions of that tool. In the next ten years we're going to see an explosion of tools that anybody can use for their risk assessment. We will literally be pulling a toolkit together of what works for your company.