Insurance companies are all about money, aren’t they? Well, yes, but they are also about something underpinning that - data. And when it comes to re-insurance – providing insurance services for insurance companies – managing and exploiting data, often on a global scale, lies right at the very heart of all the subsequent processes that occur. According to Matthias Fischer, Product Area Lead for Data Corpus and Data Foundation at Swiss Re:
When you think of these very big risks, which go into the billions of dollars of losses, that's when it is crucial that there is re-insurance. That risk, and the amount of losses, is distributed on a global scale because no insurance company alone will be able to bear that loss.
This, he goes on, is why the company is working with TIBCO to turn itself into what he calls a “data-connected company, dedicated to data collection, collation and exploitation operations on an advanced and global scale”.
As well as providing re-insurance services for the insurance marketplace, Swiss Re also works with other companies to provide them with dedicated solutions for their risk landscapes, so they can cover as much risk as possible. The third important string to its bow is helping other insurance companies to launch new products and bring them to market.
Some four years on, collecting knowledge – and not just info – about subjects is a permanent primary goals. This need has earned the company the title of `the company of a thousand professions’. This is because it employs experts for almost anything that involves a discernable and measurable risk, such as cybersecurity, earthquakes, construction, engineering, health, life and mortality. All and more have expert representation in the company.
The job of such experts is to help the company in its aspiration to make the world more resilient, so that whenever something happens to some organization’s status quo that causes a negative financial impact, the status quo can be re-established as quickly as possible, and ideally, completely. Fischer readily acknowledges it to be a `big, big challenge’, and far easier said than done.
Disaster to the left of us, catastrophe to the right
The Swiss Re estimate is that only a quarter of the known global risks are actually insured. So 75% of risk is not insured, leaving a massive amounts uncovered. This is the company’s prime target. Uninsured risks include the apparently growing number of severe natural disasters, such as last year’s major hurricane, Ian, which proved highly destructive to the Caribbean. The on-going global rash of forest fires is another obvious target.
Fischer sees such natural disasters as particularly good examples of the real issue with trying to manage such risks. Knowing, for example, the underlying risk for wildfires is essential in the future to predict where the next wildfire might happen. Similarly with floods. Earthquakes are seen as a harder target for prediction.
From an insurance point of view, the common ground here is that all of these catastrophes can generate massive amounts of submissions and claims requests. Swiss Re then uses the insights and the models it has built on its highly-curated, high-quality data assets to triage an event so that claims handlers can assist underwriters by setting the right focus.
One important issue is that the risk landscape is not steady, but is constantly changing not only through technology advancements, but also the adoption of new technologies. For example, one issue that should be higher up the agenda than it currently lies is algorithmic bias. The company’s solution to this is explainable algorithms based on generative AI technologies, says Fischer:
This is really a black box in terms of how the mind works. It bears a lot of opportunities, but also a lot of risks. And if just some of the outcome from the output is used in the wrong way, this can have disastrous effects, which we might not even think of yet. That's why we do research in this field, to be prepared, but also to share this knowledge in the community in the insurance world.
Another important source of data is that captured through wearable devices. A significant percentage of the world’s population now wears them, and even smartphones can capture some of a user’s activity data. When available with user consent, this can help build more targeted and personalised insurance products that can match the way activity profiles, habits and lifestyles change over time. Currently, the range of suitable insurance policies is pretty limited, and often become increasingly inappropriate over time for the individual covered, explains Fischer:
Since we decided to become a Data Connector company, harnessing and leveraging the data to create these insights, further advance products, and understand the change in the risk landscape, sounds easy, but it is very difficult. But we have a very good track record here. We have three reasons that makes this ambition successful.
First on his list is top-down buy-in from the Board of Directors and senior executives. Second is having the right technology in place, while third is the company’s strategy – based on the objective of creating a framework around all these many topics that make up the data it has available, all of which are equally relevant. All of them need to be advanced simultaneously in order to generate the depth and breadth of insights the business, and its customers, require. That framework is in turn based on technologies and tools providing Master Data Management, data lifecycle management, warehousing, data mesh, architectural topics, and governance.
What is important, but never works?
Fischer sees the last point, governance, as one of the hardest parts to realize:
Governance is always important, but never works. So it is a big, big challenge to overcome, for no one wants to be a data owner. And yet, it's so important. You can have the best solutions, the best platforms, but if no one uses it well it is a wasted investment.
More than a third of Swiss Re’s 15,000 or so employees are now regularly using the connected data assets that have so far been created as part of the strategy, and that process will continue while there is data still to be collected and analysed. Currently, it holds some 18,000 curated data assets, six Petabytes of data, and millions of connections between the different data options. This has meant the company has worked with TIBCO, amongst others, to build the tools to bring all that data together. This includes working with many hundreds of different sources as well as such tools as data warehouses and Master Data Management services, to build a big data platform.
Connected into this are analytics data models, plus models for the Business Standard, plus how the data is structured, governed and owned. Very strict rules are applied whenever there is a new analytics use case, with the data being taken from the ADM as a single version of truth. There is no other source allowed unless some new, unvalidated experimental data is being used before it is integrated into the data platform. If this new data is found to be of use – and in particular re-use in other models – it has to be played back into the ATM. Fischer explains:
With this we prevent the cyber-generation of copies of data in Excel sheets and other little databases. That is just a pain and keeps us from getting to this ambition of a data connected company. Also here, obviously, the top down support was a very instrumental way to ensure that we have proper governance in place properly, with properly defined standards and controls. So it's also an efficiency play, obviously, in terms of our own operations.
The ambition now is to continue growing the creation of new risk insights at scale and at speed, so that insurance companies can find that all the `heavy lifting’ of risk assessment is done for them.