How are lead underwriters doing in terms of use of technology, compared to “fast follow” vehicles?
Paul Brand, CEO of Convex: In all honesty, I think both groups are doing fairly badly. One of the key uses of technology should be to drive operational expense (opex) reductions. Lloyd’s opex seems to be stuck at 11%-12%, despite the introduction of fast-follow vehicles and five years of price increases. If you think about the price rises, that’s essentially saying opex is going up, so I’d say there’s little for the market to show for its efforts.
Sven Althoff, board member, Hannover Re: Automated fast-follow vehicles normally still operate with a human element in the loop and rely on a lead underwriter’s in-depth assessment of the business opportunity. The different starting positions can lead to different applications of technology. The most difficult part with fast-changing new technology nowadays is choosing the right solution at the right time. In some cases, it is an advantageous to be patient until the technology is more mature and, in the meantime, to prepare internal processes for the anticipated scenario.
Jason Howard, president, international, Acrisure: As regards the fast-follow model, algorithmic technology is used to establish whether a risk fits into the desired portfolio. In the subscription market setting, once the contract terms have been set by the lead underwriter, the fast-follow markets can quickly fill out the placement if the risk fits in their appetite spectrum. The fast-follow markets have developed cutting edge technology to manage their portfolios, but still very much rely both on the quality of the input data and the skills, knowledge and experience of the lead markets, the best of whom have many years of experience and knowledge that gives them their edge.
David Govrin, group president and CUO SiriusPoint: A fast-follow vehicle can be an attractive model and approach given low expense. The real question is, are the companies who are spending the money to be ahead of the charge and own the underwriting, spending their money intelligently? Is the performance compared to the expenditure worth it? Do they have more franchise value by leading and being closer to distribution and customers? For example, if a company spends 5% more to be a lead underwriter rather than following the market, are they getting more than a 5% differential in value?
Hatem Jabsheh, group chief operating officer, IGI: Neither, in my opinion, is truly utilising technology well. In contrast to fast-follow vehicles, which treat insurance as a commodity and just consider if the pricing makes sense based on their own pricing models, leaders clearly devote more resources to the negotiation process and thorough analysis of the risk. In terms of technology, I perceive little variation.
The issue lies in the wrong way the industry is addressing technology – too many programs and applications are being pushed onto the market. The best course of action would be to release the data in a standardised manner and let the market assimilate and process it.
Gianfranco Lot, CUO, P&C reinsurance, Swiss Re: We see technology playing an increasingly important role in both camps – lead underwriters and fast-followers. The focus is different and can also shift within a company depending on the strategy: while lead underwriters use technology to augment the underwriting process on complex risks – think wording analysis, submission analysis and underlying trend assessment – followers use technology to assess quickly if a risk fits within a pre-defined box trusting that the leader has covered all other aspects.
James Slaughter, chief underwiring officer, Apollo: At Apollo we are using our Smart Follow offering as an innovation spearhead. One of our key philosophies in our Smart Follow programme is that the technology must be transferable to lead underwriters. That isn't to say that the technology is identical for all underwriters however: much of a risk's digital journey will be the same (e.g. ingestions, enrichment, scoring and pricing), but the interaction with the human underwriter and the timeline will vary.
Where do technology and AI offer most advantage to the underwriter?
Thomas Blunck, board member, Munich Re: Some of the daily underwriting work is rapidly changing, as technology is taking over more and more manual tasks, like data cleansing and handling. So, underwriters’ working environment is becoming more efficient and they can focus more on carving out the various aspects of a risk on their desks and on structuring bespoke solutions.
The technological advances make new types of information easily accessible to underwriters by digging through and summarising huge amounts of documents with the help of AI, and also by identifying emerging developments much faster. This additional information improves risk assessment and the evaluation of a risk’s quality, helping underwriters to make evidence-based decisions.
Brand: Technology and AI present opportunities to speed up response times to clients, and in general, are becoming more client focussed. Clients are the only people who put money into the system so anything we can do to reduce expenses, give greater certainty on claims outcomes, maintain certainty and consistency of appetite, and simplify and speed processes will make them more popular. It’s interesting to me how clients get lost when we talk about technology. Most of the excitement and activity seems to be about making underwriters lives easier and, through additional analysis, giving an illusory sense of greater certainty about individual risk selection. My sense is this is maybe 180 degrees wrong.
Jabsheh: Pricing can benefit from AI. You can be sure the data is standardised if you incorporate it into a pricing software. AI can generate predictive analytics to assist the underwriter with actuarial and risk exposure calculations, as well as with the company's capital models.
It won’t support operations in commercial insurance. It's a different story with retail insurance, but in commercial insurance, underwriting and reserving are more important than efficiency-boosting measures. But it’s important to understand that technology just serves you – it doesn't fix your fundamental issues. When technology is introduced without a foundation, its advantages are restricted, and it becomes costly.
Slaughter: The biggest opportunity now is in productivity and the ability to scale, allowing expert underwriters to support faster, better decision-making through efficient processes and providing deeper insights. To modernise, companies can follow four key stages, using various technologies and AI tools at each step: digitisation, integration, automation and optimisation.
Digitisation converts documents into digital data under standard definitions and involves multiple disciplines to extract data from documents. Integration involves connecting systems so data flows end-to-end without integrity loss and is easily accessible – this involves extensive infrastructure work. A digitised and integrated platform enables automation with AI models continuously learning from new data. This allows advanced optimisation with ML models that enhance decision making processes in underwriting or claims.
Govrin: In terms of contract language and coverage, AI has the potential to offer enormous advantages. It helps you see if your contracts are too broad, if you are being consistent in what you are covering and do not want to cover. AI can also provide underwriters with the potential for data mining, providing information quickly on risks, trends, and supporting financial analytics. Underwriters can use it to go through reports to select themes and make comparisons quickly and increase efficiency. Using AI to look at your data bank helps you consider emerging risks by looking at historic contracts to figure out where you could be exposed. We are only at the beginning of the potential value creation.
Howard: Distribution, principally in the hands of the brokers, has used technology to provide better solutions and service to clients. AI can be used to search out new prospects, with a broad understanding of their concerns and requirements before direct contact has been made. Underwriters that can partner with their distribution to match the clients’ needs to effective coverages, allowing them to benefit from the AI investment that has been made by the distributors.
And how significant is the advantage in underwriting vs other parts of the insurance business such as claims/operations?
Brand: I think the opportunity is at least as big in claims, ceded and portfolio resilience as it is in underwriting. Particularly thinking about claims, it’s easy to imagine ways that we can significantly improve client outcomes and hopefully shift some of the negative perceptions of the industry. It’s more than a little ironic that carriers, who have failed in recent times to impress their shareholders given their poor returns, are also getting it in the neck from their clients for being too expensive, restrictive in coverage, slow and unfair in claims resolutions.
Lot: While we see the entire value chain benefitting, the biggest advantages so far have emerged in claims and operations. Here we've seen real successes, in the area of helplines, in highlighting exceptions or in fraud detection, for example.
When using AI in the right way, it makes the claims process significantly more efficient, for both insurers and their customers, ultimately benefiting both. Take for example Swiss Re's Rapid Damage Assessment platform – it uses our proprietary natural catastrophe models, imagery, weather, and property data, enhanced with AI algorithms, to assess damage at each insured property. Aggregating this at portfolio and geographic levels provides insights for faster, smarter claims decisions during nat cat events, as demonstrated during Hurricane Ian in Florida, 2022.
Jabsheh: AI may identify fraud and abnormalities in claims by applying it to standardised data from you or third parties.
Drones and AI are also compatible; consider using AI to map areas susceptible to earthquakes. However, standardising the data we use is necessary before integrating it into the ecosystem. While AI is currently all the rage, standardising data is the problem that no one wants to confront.
Howard: The ineffective application of data and technology has kept the industry reliant on the instincts and gut feel of the underwriter, which has over the years proven to lead to suboptimal decision making and overconfidence. Reducing these biases in underwriting is the strongest way of improving risk adjusted underwriting results.
The effect in operations can be significant but is applied to the expense side of the insurance P&L account, which accounts for a smaller share of the insurance companies’ economics. Nonetheless, this area requires a more collaborative industry approach to create an ecosystem that everyone can participate in that works more effectively.
Blunck: The efficiency gains with new technologies can already be felt in all parts of the organisation, but particularly when it comes to repetitive tasks or highly standardised segments. The best example is the accounting of health insurance, where automated text recognition is becoming standard. More qualitative and therefore significant changes can be seen in tasks that rely on rather unstructured information. For instance, claims operations benefit from the earlier recognition of loss occurrences.
Slaughter: At Apollo we firmly believe that the advance of (Gen) AI and associated technologies is as valuable in the claims field as other areas. However, there are significant data challenges there that will need to be addressed before the technology can realise its full value.
What share of the market will be occupied by fast-follow in future? And how can the market best protect lead underwriting skills while developing technology-based fast-follow?
Govrin: This is like asking what percentage of the equity markets should be actively managed versus exchange traded indices. There are always going to be people who are good stock pickers while some are more comfortable tracking the market. They are two different business models – and the same goes for lead underwriters and fast-follows.
Lead underwriters are closer to clients and have more franchise value. A fast-follow is going to be a macro capital allocator, so you’ll have a different franchise value when you are using lead underwriting skills and getting paid to be a good stock picker, versus paying a fee for someone else to be a good stock picker.
If you are a fast-follow, you need a quite different expense structure. You cannot have the same structures as a lead underwriter, otherwise you’ll be run out of business. A fast-follow must have a low-cost efficient expense structure while the lead has a higher expense structure.
Brand: I think it’s too early to say. So far, fast-follow hasn’t had to cope with a soft market. It’s not impossible to imagine a world where increased acquisition costs and difficulty accessing the best business more than offsets their potential opex advantage but I could also see the opposite might be true. I suppose my nightmare scenario is London’s inability to develop successful technology strategies and use those to get closer to its clients, [or to] resolve some of the opex issues which led to substantial outflows of business, particularly in the high-volume areas.
Althoff: On the reinsurance side we are looking at pools of risks which are relatively heterogenous. The ceding companies expect a differentiated approach particularly from their lead reinsurers, based on their own merits and across the entire spectrum of their cessions, rather than only contract by contract. This is supporting more of a relationship approach which still requires a significant human element. On the other hand, when looking at more homogeneous pools of risks, automated underwriting may well be one of your competitive differentiators. The optimal proportion of automated fast-followers will ultimately be determined by their success, especially in soft markets or when the impact of large losses is being felt.
Lot: Ultimately that is a decision the market needs to make. AI will only be as good as the people who develop and use it, so there will always remain a need to have experienced underwriters. Actually, we think the job profile for underwriters who write lead lines will become even more interesting. In my view, the future will be less about protecting the lead skills, but enhancing and elevating them. As tasks like data scrubbing or checking basic terms disappear, we create time and space to strengthen underwriters' skills on holistic underwriting, entrepreneurial mindset and portfolio management.