Bevis Tetlow returned to his roots in technology earlier this year, joining Imaginera as CEO, having spent 13 years as a reinsurance underwriter for Hiscox. Prior to leaving Hiscox, Bevis was chair of North American underwriting for Hiscox and Bermuda, as well as Head of Underwriting Systems.
The debate continues about whether insurers can evolve and innovate on their own, or will they become increasingly reliant on third parties for new ideas, or perhaps even disappear? Matthew talks to a former colleague from the early days of catastrophe modelling about what reinsurers really need, how to give it to them and what blind spots still remain for property reinsurers. Bevis reveals that despite it being 30 years since catastrophe models first emerged there are still lots of opportunities for companies that have data to solve problems for re/insurers that the models still can't solve - fraud being one of these.
Bevis signed up bespoke software developer Imaginera to build the reinsurance underwriting system for Hiscox. He has now hung up his underwriting pen to join founder Toby Crawford in expanding Imaginera's business in Bermuda and the UK. More information on Imaginera here - http://www.imaginera.co.uk/
Transcript for this podcast
00:00 Bevis Tetlow: I think your job as an underwriter is to basically piece together all the different bits of the puzzle...
00:11 Matthew Grant: Welcome to the InsTech London podcast. This is Matthew Grant and this week I'm talking to Bevis Tetlow, CEO of Imaginera. Bevis started off his career in catastrophe modeling, launching his own technology company, then moving into underwriting with Hiscox. He headed up their North American treaty insurance for their Bermudan office and whilst he was there he designed the company's reinsurance underwriting system. This was subsequently built with the help of Imaginera, and Bevis joined the company earlier this year. We're delighted to bring you this podcast with the support of Insurance Insider. We know from experience that Insurance Insider readers are the first to hear about the latest global wholesale specialty and reinsurance breaking news. InsTech London listeners are eligible to download the latest issue of Insurance Insider free from insuranceinsider.cominstechlondon, all one word, and we'll give you that in these episode notes.
01:13 MG: Bevis, welcome to the InsTech London podcast. You've had a pretty varied career. We started off together at RMS about 20 years ago. You've gone from technology, you went into underwriting and now you've come back into technology again. So, what is it about the lure of technology that brings you back?
01:31 BT: I mean, actually, it was a very hard decision to leave underwriting, but I really enjoyed my time in Bermuda and being an underwriter is a great career. Back in the day, we used to work at RMS and after that I worked... Had a little consulting firm in technology. That's always been an area been very interested in and the opportunity in Oxford to work at Imaginera was obviously appealing too. So that was part of the reason to stay in reinsurance, but not be in London.
02:00 MG: We'll come back to Imaginera in a minute, but just talking a bit about your time at Hiscox. So you were the underwriter there for US treaty business, but in your spare time, you seem to have built the entire reinsurance underwriting platform. How did you manage to combine both, what it seems like, two full time jobs?
02:16 BT: Yeah, so I used to head up the underwriting... The treaty underwriting for North America and for the Bermuda and London operations at Hiscox. I didn't build it myself at all. I obviously had a big hand in designing it, so I used to head up the underwriting systems design at Hiscox, but the truth was there's a big team of people that helped me and I was simply the... More the figurehead of it.
02:40 MG: But it seemed to survive and actually I think, is still been used pretty successfully today at Hiscox. Is that right?
02:46 BT: Oh yes, absolutely. So Hiscox has built an entire end-to-end system for themselves and I had a big part in moving them when I joined Hiscox, would have been 14 years ago, into that world away from Excel into a much more bespoke platform in terms of how they processed and managed their submissions and did their modeling.
03:11 MG: That choice to build versus buy is something that lots of companies struggle with across all dimensions and, in particular, in terms of some policy administration systems. What were the choices you had at the time when you looked at... To see what was in the market versus what you have to build yourself?
03:29 BT: Certainly back when we first went out to Bermuda in 2006, there was very limited opportunity to buy anything in. Hiscox would have preferred we'd bought at the time, but there just simply wasn't anything available that was fit for purpose. So really there was only one option, was to build it yourself. And in most big reinsurance companies have gone down that route because that becomes your IP.
03:53 MG: And you got know Imaginera as one of the providers or the builders of your system is that right?
04:00 BT: Toby, who set up Imaginera was originally in Bermuda and that's how we got to know each other and he went back to the UK to start Imaginera.
04:10 MG: Tell me bit about Imaginera themself.
04:12 BT: Toby Crawford and his wife started the company about five years ago when they moved back from Bermuda and we now have about 30 coders based in Oxford and Bristol. We have a couple of very big clients so Hiscox is obviously one of those, Fidelis another. And I think we are fairly unique in our ability to offer a sort of real world understanding of reinsurance combined with sort of coding technical excellence. We have, obviously at this point, a very large amount of experience in building end-to-end reinsurance platforms. So we sort of specialize in helping people develop their own systems which they own the IP of, but increasingly with me coming onboard the future will... We're starting to look more at what partnerships might look like and Imaginera starting to build up its own IP as well.
05:07 MG: And presumably, most companies don't make a decision fairly frequently, or they'll make a decision only every few years, even decades about which systems they're using. So how do you, as you look at growing the business, how do you map out the future when you've got really a very short window, on a fairly regular basis to find the companies to work with?
05:28 BT: Yeah, the truth is that's a real challenge and ultimately, we... What we want is fewer clients, but those clients to be very sticky and be... We're very much a partnership, a external coding source for them, so in many ways we are be part of their team, but just as an external contractor. But it's obviously, it's a big decision to start down a route and once you start down that route it tends to be very sticky. I guess with my background it's... My knowledge of how things operate at the different companies is helpful in knowing how to target.
06:02 MG: So just moving on to this issue of interoperability. So one of the challenges the market has had for a number of years is a lot of systems have been designed in silo and it's very difficult to move data from system to system. Very difficult to find some common themes across multiple databases. From your experience of building out the platform at Hiscox and also what you see now at Imaginera, do you see things getting better in how these different applications are able to talk to each other?
06:35 BT: Yeah, I think it's gradually getting better, but there's still plenty of scope for improvement and I guess these are areas which we are looking to help people get better at. There's still a lot of dysfunction in terms of how data is transferred between companies, how modeling data is moved around, how brokers communicate, mainly via email still. And I think there's a lot of room to make that easier. There's certainly examples of where you can kick off the modeling automatically with AR, RMS, for example, these days, which makes life a lot easier via APIs, but I still think there's a lot of scope to do this better and that there really it hasn't been a industry-wide standard that's been adopted universally.
07:20 MG: And as you're talking to the companies you're working with and looking what else they're using, how much information is available if you want to help them create connections between other applications? Is there a fairly good set of data out there about how the data moves between different systems?
07:37 BT: There's a lot of variation out there. Some APIs obviously work very well and others, there still needs a lot of customization to allow you to do what you want.
07:49 MG: Let's talk a bit about catastrophe modeling. So you started off your career with RMS, running models, you used them fairly regularly as an underwriter. What was your view on the state of the market for underwriting models today? Have all the models reached maturity and saturation, or do you still see opportunities for improvements in modeling?
08:11 BT: Yeah, I think there's still a lot of areas that could be improved to help the underwriter get the total answer. And I think in terms of things like the standard and more mature models such as the US Hurricane model, an earthquake model, you know things that at this point are probably as good as they need to be. But there's a lot of room to do things like the non-model perils better, there's still lots of missing perils like flood. But increasingly, I think the answer is, as an ex-underwriter trying to get to the total answer, includes things like, how well does the company operate, how does it deal with fraud? When you've got a situation, Michael, where 25% of the loss is coming from fraud, how do you think about that as an underwriter, and how do you measure that? I think that comes back to big data sets and how you interact with those big data sets. So you need systems that will enable you to bring all those different answers together from lots of different various places and consume large amounts of data to give you a sensible answer at the end of the day. So it's a challenge and I think there's still plenty of room to address the other side to modeling.
09:26 MG: So there's still lots of options for data sets but not necessarily what maybe traditionally would have been thought of as data for modeling, such as hazard, but more about human behaviour and cost of rebuild and other areas outside of just the peril itself.
09:41 BT: Yeah, absolutely. I think fraud, understanding fraud is a very big and important topic and I think big data can give you some pretty strong answers in that. Different people write different sets of portfolios, you have client A, who has a very mature elderly client base middle income, they're going to respond very differently to a younger Miami-based set.
10:06 MG: A question back to your underwriting days and just reflecting on the fact that there are a lot of companies now coming in, offering different types of data to help people make underwriting decisions or make better underwriting decisions. When you were an underwriter, which wasn't that long ago, what kind of data was it that actually really helped you make different decisions on the treaties and the portfolios you were looking at?
10:29 BT: Your job as an underwriter is to basically... To piece together all the different bits of the puzzle, to try and... Really what your interested in, is how companies compare, so the comparative measures, any data that allows you to compare companies versus another company, portfolio A versus portfolio B, based on their exposure I think is helpful. Because ultimately, your job is to pick who's doing a better job, relative to the price that you're receiving for that business. So data around data quality of buildings, such as some insurance to value for example, might be a good one to look at, relatively. So we use big data sets like that to decide areas around how much exposure we might deploy on one company versus another. So, it's... Everyone can run the standard model, it's... What the insurance companies of the future are looking for additional sources of data, which can augment their decision to allow them to better calibrate Company A versus Company B against each other.
11:38 MG: Yeah, I guess it's a very important distinction, because a lot of the discussion around data relates to pricing or risk selection but actually what you're saying is that it was really a question of how you deploy your capital, your capacity on the different businesses that you write. So you make sure you're kind of maximizing your capital on the ones that you've got the greatest confidence in, and the other ones you might still underwrite them but you might just be a little bit more severe on how much of their capital you allocate.
12:08 BT: Yes, and to the point where you know, a company that does a really good job, actually deserves, potentially can get a discount. So you're prepared to write a reinsurance contract at slightly less, if you're really confident in the underlying exposure relative to, and a bigger line, relative to a company where you know that the underlying exposure is a little more shoddy, you are less likely to deploy a big line and you're going to want a bigger price.
12:36 MG: So, just... So talk a bit about how you work with people. One of the challenges for organizations is when they engage the technology company, how to specify the product, how to make clear what they want. What guidance from your own experience and what you're doing now you're going to give to people to get the best out of that technology provider when they're specifying their systems.
12:57 BT: Yeah, the truth is it's not an easy thing to do well. And each situation tends to be slightly different, but I certainly prefer a much more agile approach. The way we work is, I think, because we come with a lot of prior experience it's not you explaining to a BA that then... To a business analyst that then moves that on to a remote developer. We're coming with a real understanding of what most people are doing day to day, and so that proximity between the customer, the end user, and the developer, when the developer really understands what the end-user wants, is the best approach, I believe. And you can develop things very cheaply and very quickly that way. And what we tend to do is work in two week sprints where we... The user is able to help understand what they've... What direction they're going in and also change direction if they want to.
13:56 MG: Bevis, you've been in technology for 20 years, you've also been there as a practitioner and an underwriter. About four years ago the term Insurtech popped up as a whole bunch of companies arrive claiming they could change insurance. What's your perspective on where we are now?
14:12 BT: Yeah. Despite being a massive fan of technology, obviously with my role, I'm a little bit skeptical. I've seen a lot of people come and go. I do think there's lots of room for efficiencies, but personally I think that the big successes will likely be with the existing players getting better and becoming more efficient rather than a big revolution of new start-ups that are going to change the world.
14:40 BT: Take, for example, a company like Lemonade. Underneath the surface they're really just an insurance company that sells renter's insurance in New York. I would say they've done a fantastic job of marketing, but in many ways they are an insurance company much like any other. So I believe that there's lots of scope for the existing players to improve and we, obviously, hope to help them do that.
15:11 MG: Good. Well, you can be a part of the innovation journey. And so on that, anybody that wants to find out a bit more about Imaginera or track you down, what's the best way to find you?
15:22 BT: Oh, yes, certainly. Well, obviously the website's an easy way to start. So, Bevis Tetlow is my name and Imaginera is the company.
15:32 MG: Good. Well, Bevis, thanks very much for carving some time out to come and catch up.
15:35 BT: Thanks very much. Bye-bye.
15:42 MG: In the four years since InsTech London was founded, we have spoken to and heard from over 300 companies driving innovation either as buyers or providers of technology at our monthly evening events. You can hear most of our speakers for the last 12 months on this, the InsTech London podcast. And further information and details are at our website at www.instech.london, including registration details for our forthcoming events. We are supported by close to 100 corporate members, including Insurance Insider, and if you'd like a free copy of their latest issue you can find that at insuranceinsider.cominstechlondon. And details will be on the episode notes.