This a shortened version of Making Risk Flow podcast, episode: “How to Unlock Higher Returns on Investments in Analytics”. In this episode of Making Risk Flow, Juan is joined by analytics expert Paul Mang, Innovation Advisor at Guidewire and former global CEO of Analytics at Aon, to discuss the use of analytics in insurance and how to get better returns on such investment. Juan and Paul also address how the industry can make progress through interoperability, ways in which analytics can improve efficiency whilst driving down costs, and why the insurance industry needs to focus on increasing its relevancy.
Listen to the full episode here
Juan de Castro: Welcome to another episode of Making Risk Flow. In today’s episode, we’re going to discuss how to unlock higher returns on our investments in analytics and to discuss this topic, I’m joined by one of the global experts in this area, Paul Mang. Paul, Thank you so much for joining me and welcome to the podcast!
Paul Mang: Juan, thanks for having me. I’m happy to be here. Always happy to talk about one of my favourite subjects: analytics and insurance.
Juan de Castro: Absolutely. So, for those people who don’t know you, an overview of your background, please.
Paul Mang: I am currently at Guidewire Software and lead our innovation efforts primarily working with our largest global clients on their innovation efforts. Prior to that, I was the global CEO of Analytics for Aon mostly working with our analytics team at the enterprise level, our actuaries on the reinsurance side, CAT modellers, building up our own centre of excellence at Aon around analytics. And prior to that, spent probably the majority of my career, I was a partner at McKinsey and one of the leaders of the P&C practice there in Chicago and in London.
Juan de Castro: It’s quite an impressive background. Let’s start with the obvious question on the topic, which is, why should we focus on better returns on the investments in analytics?
Paul Mang: Look, I think that’s an important issue. And at one level, it’s very straightforward. Clearly, if you’re using analytics, you want to have a good return return. If you’re selling or serving the industry, you want to have your customers get a good return, and I think that’s important. But if you think about things at the industry level, I think our major challenge within the insurance sector is relevance. Relevance to business and society, how we as a collective sector address volatility, risk, and apply capital to it. And I think we’re a bit stuck right now, not that it isn’t a healthy industry, but as a relative percentage of the economy, the global economy, we are shrinking in share, and all of us have seen things written, podcasts, interviews about the protection gap, that in fact, as the world in some places gets more volatile, we haven’t really matched the solution so that we could cover more and more risk. So I think if we are going to challenge the big problem that the industry faces, analytics is going to have to be a big part of that. And to get more of anything, to just do more of anything, to invest more, to have people invest in creating new solutions around analytics, we need a better return on those investments and analytics. And I see that as the positive flywheel. If we can do better and get more return on the analytics we have, it will improve the economics, the relevance of the industry. It’ll also be an incentive to generate more investment, to have more innovation and analytics. So I see that as solving one of our big macro problems.
Juan de Castro: Is that because analytics will help us identify those opportunities to become more relevant? Is it because analytics will also identify opportunities to drive a more efficient and effective industry and therefore a more affordable one that can be more relevant? Is it the combination of both?
Paul Mang: I think it in fact does cover all of that. I think about it in two general categories. One is around efficiency. The protection gap largely has to do with the price of covering exposure. And in fact, I’ve said in different venues that the biggest competition any of us face in this industry really is not necessarily incumbents competing with other incumbents, or in fact incumbents as a group competing with startups, with disruptors. I think the biggest competitive issue facing anyone who participates in this industry is non-consumption. In fact, there’s more business not being written. And so it’s very hard to take away business from a competitor if you think about the alternative, which is your key competitor is potential insureds are self-insuring. They’re either retaining it because of higher retentions or in fact they are becoming their own insurer by applying their own balance sheet against volatility. So that’s one thing is analytics could help improve efficiency. Efficiency should drive down costs. I think we should have a target to reduce every, if as an industry over many decades we’ve returned about 50 cents on the dollar. If you just look at loss ratio, so every dollar premium coming into the industry, we’re returning back to the insured’s collectively about 50 cents. We need to do better. And if we don’t do better, then that is high enough cost where insurers will be themselves being their own insurance entity and doing it on their own balance sheet. And by the way, companies now are larger and larger. So the largest private sector companies, they have market caps of trillions of dollars. If you think about the size of the insurers, they have a capital base that in fact they could substitute for us as an industry. So that’s number one. Efficiency has got to be part of this to address the protection gap. And the second is we should think about underwriting innovation, not the way venture capitalists do, which is they provide investment and they’re making in some ways an assessment of future earnings. But any new innovation really puts some risk into society. So autonomous vehicles are great until there’s a problem. And you may not have very, maybe a low frequency environment, but if you have a major problem, you may have a million accidents at one time. Innovation is pushing us into high severity environments. And I think we need analytics to address those emerging risks. And so when I say we need to underwrite innovation, I don’t mean we need to act as VCs. We need to think about the kind of volatility associated with innovation that we critically need as a society. I just listened to a session about financing climate technologies. These are climate change and the dangers associated with severe weather. It’s one of the big problems of the day. And the technologies that scientists are coming up with to address it, it’s amazing. But some of those technologies are a little bit scary, like carbon capture, right? You know, affecting oceans and new kinds of food. And these are spectacular innovations. And just as an insurance person, you think about what happens if something goes wrong. And so there’s a lot of discussion about how to finance these new innovations. These things that we need as a society. And I think we as the insurance sector need to think about how do we manage the risks? In fact, if we can’t come up with solutions to manage those risks, the financing won’t be there. I don’t mean to reduce the risk of the initiative failing. That’s what the VCs and private equity firms do. I mean the other unintended kind of a little bit of the unknown and there’s a bit of innovation here. And so I think there’s a whole second area that has little to do with efficiency, but it’s still at its core is analytics and allow us, if we can do these two things Juan, I think we will have made a lot of progress on relevance. We will have taken our rightful place being the insurer rather than pushing risks onto the balance sheets of the insureds who are busy doing other things. They’re inventing new things, they’re bringing food to the table, they’re generating energy, they’re transporting. Why is it that they’re acting as the insurance sector whereas this is supposed to be the business we’re in? And then all the new innovations that are needed in society, they need some sort of risk transfer, risk management to state it more broadly. I think both of these things could be or should be the objective of investments in analytics. And getting the better ROI on the analytics we have right now is the path for our industry to do that. It’s a long way of what motivates me and focusing on the ROI on analytics because I think it serves one of the major objectives we should have, and that is to really increase the relevance of the sector.
Juan de Castro: In order to create relevancy, you touched on two areas. It’s making insurance more affordable by driving efficiency in the industry, and by driving innovation in the way we manage new risks, new exposures. I had never heard about it that way, and I think it makes complete sense. One of the topics you’re really big on, and it’s one of your mantras, is that interoperability is the key to unlocking this value. Can you tell us a bit more about what you mean by interoperability?
Paul Mang: Interoperability is the reuse of some data or models so that you could actually get, you can get a return, like perhaps two positive outcomes for the cost of one. So interoperability allows the sharing, one element of interoperability is sharing. In fact, someone I’ve talked with recently said, you know, what you’re talking about sounds like conservation Paul, sounds like you want to reuse, recycle, and reduce. I think that’s the mantra of conservation. Reduce what you consume, recycle what you can, and reuse. And I don’t think that’s a good analogy. Because for example, you could imagine then the same data being acquired, purchased by the same primary carrier to do their underwriting. And then maybe it again gets purchased later on during the claims if there is a future claim and then it gets purchased by the reinsurance broker, same assessment of physical characteristics and then the reinsurer. So in this fanciful example, the same data, although it could be purchased at different points in time, could be repurchased multiple times. Now the ROI on any one buyer is whatever it is. If we had a mechanism, I’m not talking about standards, but a way to make it more interoperable so some of the insights could slide down that value chain. I think fundamentally, we’re reusing the data, we’re doing it in a way where the industry would be better able to allocate capital. And so that’s why interoperability, I think, there’s, again, I’m not saying it’s the only way to help improve the returns on investments to analytics, but thinking about how partners, transaction partners, could share in some of the insights, still within the bounds of competitiveness, but how do we share some of the data, some of the insights, some of the models, and have it flow through that value chain. That’s the next step of getting better returns. We’re working on all the basic elements of just implementing a solution at one point in time to affect one decision, but can we have it also affect the other decisions in the value chain?
Juan de Castro: Almost like being a bit sceptical, you would say, well, why would you care? This is why the introduction you made about relevancy is vital. Which is definitely if each of the actors in the value chain purchases this data, they will be less efficient. The whole value chain will be less efficient and the industry will be less relevant.
Paul Mang: I think what you said is absolutely right, but I don’t think it requires you to have your perspective on the unit of analysis at the industry level. I think even if you take the point of view of a single participant. Imagine the benefits if you implement solutions within one of your silos. So you do some analysis, gain some insight within claims, but you don’t share it with underwriting. So you actually determine something in that part of your within your four walls value chain. If you don’t share it or don’t make use of it on upfront underwriting or even further how you market and select, you can see you’re leaving some value on the table. And in the same way, you could imagine if you could share insights across the value chain outside of your boundaries, it isn’t just at the level the whole industry gets more efficient, which I think it would, but you, yourself, the entity, the protagonist of this example, will gain benefits. We’ll get both. It’s almost like having your cake and eating it too. You’ll get the benefits, the industry will get the benefits, but also it’ll accrue to the one decision maker. At least that’s my hypothesis. We do have examples of interoperability. And so we’ve got multiple CAT models out there. There are some similarities, some differences. They are shared across. The way we think about risks and exposures, it’s through the lens of the newly developed CAT models. And we have other examples. Accord provides forms for small businesses and other lines of business. And they’re not used universally. I mean, Accord forms are used, there’s a certain percentage used within the marketplace. And so we’ve already used examples. There are already examples where they’re sharing of certain data across multiple parts of the value chain. We’re nowhere near as far along on this element as we are within the boundaries. I think we’re working really hard to share the breakdown silos, breakdown boundaries. I know Cytora’s focused very much on working through that entire workflow and sharing insights. So I think there’s a lot of good work going on within the boundaries. I’m only trying to highlight some of the opportunities, if you think, of opportunities outside your own organization.
Juan de Castro: Ultimately, you’re talking about making risk flow across the industry. You touched on a couple of examples, or like Accord. And often when people think about interoperability, they think about standards. Is it the same, or is standard just a subset?
Paul Mang: What we think of as standards or standardisation as the strong form, the strongest form of interoperability where you agree ahead of time. It could be by convention or by regulation. There could be a reason why we all agree to a certain standard. But I think all of us and all of your listeners to the podcast will know that we don’t have a great record on standards. We’ve been trying to have standards and how we submit data. SEMC and Lloyd’s is famously working on and has worked on for years, standards on placement. And there are reasons, I think all of us, there’s no need to get into details, I think all of us know some of the challenges of standards. There’s a political, strategic, whose standard will it be? And importantly, does standardisation limit innovation? Because once you start locking in things into a standard, there are those in the thought leaders who have observed that’s one good way to lock down and in fact be a barrier to innovation. So to me, interoperability is the weak form. Most of what we’re talking about is the weaker form where there’s enough commonality where you could imagine someone on the other side, so a different part of the value chain could and could understand and gain insights. But we’re not forcing everyone to adjust their own internal operations, their own technology architecture to fit a particular tight coupling. But it requires a loose coupling. We do it all the time, imagine insurance back a century ago where we didn’t have the technology that we have at our fingertips now. You know, an underwriter might rely on using just verbal language or mail. They would understand a risk, but they would have to have some understanding of the other person, the other side of the transaction, understand their reputation, understand their capability to be able to interpret if a broker described the risk in a particular way. Using verbal communication perhaps, there was some transmission of information. And we did it through relationships, we did it through a shared understanding of some terms. We didn’t force everyone to speak in exactly the same way. We relied on reputation and other conventions. Well, today we’re using technology and it’s hard to look at a spreadsheet or look at a stream of data and understand the reputation, how correct, how accurate, how much data integrity does this flow of data have because there’s not a person behind it anymore. But to me, when we’re in a world of interoperability, we will be making assessments that this data source has more integrity because I know from history that it has evolved in this way. And if I’m looking at a different data source and trying to fit it into my own internal models, which again have not been forced to a position of standardisation, I have an understanding of how I could fit. I have an understanding of insights that came from my transaction partner and I can make use of it. If I don’t have that, then everything looks very messy. In fact, that’s what I hear most when I’m talking to clients or organisations. I have a lot of data, but it’s all messy. It’s hard to use all the data I have inside my four walls. And then I’m getting data from new special, innovative data providers. It’s coming at me. It’s hard. I need to tweak it because it’s changing, the risks are changing. So we’re not in a situation where we’re lacking data. We’re in a situation where it’s hard to interpret and understand it, and I think interoperability gives us the path, that understanding, the shared, the commonality. It is the way for us to reuse data and do better with the analytics and data that we have.
Juan de Castro: What have you seen in the last five, ten years where have we made progress and where are we still behind?
Paul Mang: Well, I think one example of, in fact I had my own event like this one where I interviewed some people and one example made an impact. It was a couple of years ago, it was during the pandemic period, I had Ian Branigan at Renaissance Re join me and we talked about one example that they had where they had invested a lot in flood models. But if you know Renaissance, they’re not in the direct side of the business, but they had what they thought was good enough in their insights on flood. They felt comfortable to push that model down to an agent. This is typically not what they would do, but they pushed that understanding because there weren’t very many alternatives. And What they did was create a business. At first, it was just, I assume it was a pilot, something smaller scale that grew. But they put those models into the hands of the retail agent, a subset of them, and they were able to provide a better product to those insured. So essentially, business that would have gone to the NFIP, the National Flood Insurance Program in the U.S., instead some of that business would go to Renaissance book. And I think that was a small but important example of an investment in analytics done at Renaissance because clearly they had an objective of understanding a portfolio, that’s what you do if you’re a reinsurer, you look at portfolios. That provided enough insight where they thought they could push that further up the value chain, push that to the agent. Remember, they’re not looking at a portfolio, they’re looking at business one at a time as it comes in. And again, that’s the mental shift that has to happen. They invested in portfolio assessment, but they were able to turn it on its side, make it useful for a decision maker who’s looking at everything risk at a time. I think it’s a great example of conservation, of reuse or recycle, I don’t know which one it is, but they’re reusing their insight and that makes that risk flow better. They were able to create.
Juan de Castro: It’s about reducing the number of actors that are just generating the same insights. So if you had to say, what are a couple of areas where the industry still needs to make progress in this area of interoperability, where would you focus?
Paul Mang: I think this last example is a good place, the kinds of investment and the work that the reinsurance subsector have made, I think some of, not all, but some of those models, some of that analysis could be in fact helpful in the primary. Here’s a future that I think we should be, it’s hard for me to say we’re heading in that direction. I would like it. I think it’s a logical place for us to go. I hope it’s a direction we’re heading towards. I think there’s a big opportunity there. The simple way to describe it is if you’re a primary underwriter, you’re looking at risks one at a time because that’s the way it’s coming to you. I mean, there are exceptions to that where you’re looking at a whole book or a program, but you’re looking at risk primarily one at a time. Now you do have guidelines and you do have an appetite guide. You’re not writing everything independently. You have a strategy of where and what, but you are still making a decision at the risk level. I can imagine a future where you’re not just waiting for that book to be created during the year and then you’re looking at your capital management. You’re now asking the question, what do I do so I have the right capital structure based on the portfolio I’ve just written? So I’ve just spent a year writing business. Now I’m looking at the portfolio. Of course, now I’m using my advisor, who might be a reinsurance broker. I’m talking to reinsurers. In the future, I think we would be at a place where we could look at the individual risks with an eye towards what will the future portfolio look like? How will it contribute to the volatility of the portfolio? Now in that future state, the primary carrier would not be wondering, now I’ve got this book of business, I’ve got to think about these other decisions, capital management. I’m now thinking about it linearly. Instead they’re shaping their book. This is one line of business I’m talking about along the way. They need to think about it compared to all the other lines of business, but they’re shaping it because they know the position they will be at when they’re thinking about reinsurance potentially or how whether, because there are alternative ways to manage that capital. But I think that’s an area of opportunity. It’s a place where we’ve already invested. The industry has already invested in these models. The reinsurers have already invested. The primary carriers have already invested in risk selection and pricing models. Now it’s interoperability, I think, that unlocks that additional value. I look forward to that. We would do what we should be doing as an industry rather than having our customers build their own products. You can tell that I’m quite motivated by this because I see this as akin to, if we were in the energy generation business, however we’re generating energy and we’re selling it to households and businesses, I would take it as a very negative sign if my customers decided to all create their own energy generation plants. So instead of coming to me and saying, you know, we’re gonna sell you energy to your household. I responded to them by no, thank you. I’m building my own nuclear power plant. I think it’s more efficient for me to have my own coal fire, natural gas or hydro, whatever it is. If we were in the energy generation business, I mean, talk about reduction of relevance. If your customers are generating their own energy because you can’t, as an industry, deliver energy to me more efficiently, and I’m going to the trouble of building my own gas-fired plant or my neighbourhood, our neighbourhood is going to create our own gas-fired plant. I know that’s maybe a stretch as an analogy.
Juan de Castro: It’s definitely a good one.
Paul Mang: That’s what I see happening when customers, whether they’re personal lines, households or businesses say, I’d rather not have this insurance. I’ll just, I’ll absorb that on my own balance sheet. That for us is the same thing as an energy customer saying, no, thank you, I don’t need your energy, all the transmission lines coming to my house, I’m gonna build my own plant. I think that’s kind of an indictment on the value proposition and how we’re operating. So I would like us to get to the point where we can get to this efficiency and again, back to this point, relevance.
Juan de Castro: At the end, what you’re saying is going back to the original point about the relevancy of the industry, to drive that relevancy, we need to optimise capital allocation. Right now, it’s difficult to optimise just looking backwards. So you’re saying, okay, how can you look forward? When you’re analysing a given risk, how can you understand the implications of writing that risk into your book? In order to be able to understand the implications, you need to ensure this interoperability, and you need to enable data to flow within the insurer, so across your aggregation models and your pricing models, et cetera, but also across the value chain to understand the full implications of that.
It’s June, 2023. We need to talk for a few minutes about generative AI. So any thoughts about how the new opportunities generative AI brings to this concept of interoperability?
Paul Mang: I think it allows us to gain some of the benefits of standardisation. Back to the messy data, if you don’t have standardisation, then the biggest barrier is the quote messiness of the data. It keeps changing. It’s not in the right format. It needs to be transformed. When I was at Aon, we had a small army of people who would interpret data from various customers, AM Benfield, AM Reinsurance customers, and the models that we also licence to make sense of all of this. It was a very manual. Now, parts of it were automated, but in the end it was manual. It required knowledgeable people, who understood the history. I think the latest models allow us to learn to do this more quickly, more efficiently. So I think the advances in AI allow us to release some of the constraints, allow us to interpret the data and make use of it without forcing us into standardisation, which, again, to reiterate, I think comes at some cost. There’s a strategic cost of that where someone won’t participate because they think their standard is better, they don’t want to contribute, they don’t want to participate in an industry standard that gives some other competitor an advantage. But I think it could potentially also limit innovation. But the latest analytics will allow us to make better use of the messy data and I think get to interoperability. So I’m optimistic this is a good use, an initial use case for the industry. How can we share data within the boundary of the organisation and then across it by using these analytics tools?
Juan de Castro: Before we wrap up, I would first recommend everybody take and read a white paper you’ve recently published called: Do You Really Have a Digital Strategy? I found it really, really insightful, that paper. So before we wrap up, can you give us what are the three takeaways or the five takeaways from that document?
Paul Mang: Well, thanks for raising it. I think it gets at it, although it’s not reflected in the title, it does get at interoperability. Every organisation, I think without exception, that I now interacted with over the past two years has some sort of transformation, digital transformation. They have different names, but they’re largely trying to do the following, make better use of data, use analytics, use new tools to be more efficient, to be more innovative. They all have a set of objectives that fall into that broad statement. When I looked at and compared what each organisation, and this is across various parts of the value chain. So primary, underwriters, the brokers, the reinsurance brokers, reinsurers. You know, so what are they trying to do? And that’s why it was titled, Do You Really Have a Digital Strategy? So I think all of them have as an objective, we need to better consume data. We need to better use data that we have, and that gets to breaking down silos, better use of external data. I think that’s a common element. The second is we need to do something with this data. We need to analyse it for insight, some predictive models. So that’s mostly in all of these transformation efforts. We need to change behaviours. It’s not enough just to have data and understand it and to make predictions and to analyse it. We need to actually change behaviours, and that is to make changes in the workflow. And I think the theme of your podcast, the way risk flows, I think gets at driving decisions and making real actions that matter based on this analytics. And then one that’s common, I think to all of them is a modern interface with customers. Oftentimes it’s with a smartphone app and they need to interface with customers more effectively and the way people buy products today. And that gets to this interoperability. How is it that we can consume analytic insights, data that’s been manipulated so that it could reveal insights? How can we consume that and do it across this value chain outside? So that’s a simple checklist of the assertion that if you really have a digital transformation strategy, you would at least have some element of all five of these somehow part of the transformation. I’m not trying to say that we’re not making progress. All those entities that have transformation initiatives are making progress, but to have a comprehensive one, that’s why I’m firmly of the belief that interoperability belongs on that list and is a powerful way to get value out of all of these transformation efforts. And so you can see if you do read anyone there in your audience reads that you could I’d be interested to hear what they have to say about that perspective.
Juan de Castro: Paul, as always, it’s really fun and really insightful chatting to you. So thank you so much for joining me.
Paul Mang: It’s great to be here. Thank you for inviting me. And I do, I really do look forward to hearing from folks with their perspective on this important topic. Thanks again, Juan.
Juan de Castro: Thank you, Paul.
