Lets Talk: Alan Milroy on Big Data
Q&A with Alan Milroy, International Property Catastrophe Underwriter for XL Catlin
Alan has 25+ years as a frontline Property underwriter focusing on global companies and with an emphasis on risk management and tailored solutions. He is based in London. Since 2012, he has been responsible for our international Property catastrophe (CAT) book. In this role, Alan is charged with developing an underwriting-focused view of CAT exposures. This includes working with different underwriting teams to help them better understand what CAT models can provide, and as importantly, what their limitations are.
There has been a lot written recently about “big data” and its potential implications for the insurance industry. What does it mean for your business of assessing risks?
“In todays’ quickly evolving world, the term ‘big data’ refers to immense volumes of data flowing with increased frequency and from a widening variety of sources.
“While the term is relatively new and the concept topical, data and analytical tools have always been hugely important to the insurance industry in how we identify and understand risk. Today we think that big data presents major opportunities, not just for the insurance industry, but also for our entire economy.
“A good example of big data can be found in how water levels are monitored in the UK. Every 15 minutes information collected from stream gauges is transmitted to a central location. This data gives almost instant information on the development of floods – information we did not have before.
“Another important characteristic of big data relates to the diversity of the forms and formats of the data sources: structured information from public agencies; streamed data from sensors or meters; unstructured information from social media. At the same time, insurers accumulate a range of customer-specific information in the course of daily business like location data, financial statements, claims reports, construction blueprints and so on.
“In many ways, the opportunity for insurers to utilize big data is twofold:
“Firstly, new and better data allows underwriters to understand risk more completely so they can make better decisions about risk selection, terms and conditions and pricing. Secondly, technological changes and innovation occurring across the entire spectrum of society are changing our customers’ operations and risk profiles; big data is an essential tool when it comes to providing relevant insurance solutions to our clients today and in the future.”
In your role today, how are you using big data?
“The initial thrust was to bring together location information with hazard type data including flood maps, elevation surveys, soil maps and data on construction methods. All of these elements have an important impact on our understanding of Property risk. By developing a clearer and deeper view of risk, we aim to help our underwriters make more informed underwriting decisions, and our line-of-business managers more aware of the span of risks across a portfolio.
“For example, two years ago we looked at a road in Russia. Roads are difficult – there are a lot of different factors you have to consider like bridges, tunnels, different terrain, etc. Typically the underwriter would manually select a certain number of points along the roadway, hoping those locations accurately represented the overall risk profile along the full length of the road.
“When we first looked at this road using a traditional approach, we saw quite a lot of flood risk in certain areas and opted not to quote the business. We looked at it again a year later using a data source that gave us good information along the entire length of the road. By combining that with detailed data on how the road was engineered, we were able to visualize different sections in considerable detail. We discovered that the sections in the riskiest areas were highly elevated. So the flood risk was actually far less than we saw from a two-dimensional map. We felt quite confident at that point about our understanding of the risk, and ended up writing the business.”
How do your clients benefit from these efforts?
“Having open, fact-based discussions with our clients can only enhance the partnership. As we develop a deeper, more multi-faceted understanding of a client’s risk, those new insights can always help a client improve its overall risk management programme.
“Big data analyses can enable a client to better identify and prioritize its options for mitigating risk and creating a more resilient business. And armed with a fuller understanding of its risks, the client can more accurately structure its risk transfer elements in line with its risk profile.”
There seems to be a proliferation of data sources – in your role how do you select the data?
“There is a lot of very good data out there.
The challenge is identifying who is collecting what, and then determining how it could be useful to us.
“I would say that 90% of the data we use is from publicly available sources, particularly government agencies. We get a lot of information from the United States Geological Survey, including data on events happening anywhere in the world. The UK Environment Agency does a fantastic job putting out a variety of data, especially related to flooding and other environmental issues. And at the local level, we can now get a lot of good data from municipalities on building stock.
“We’ve also seen a number of open source communities spring up around different topics. For example, Oasis is an open source modelling community that was created to expand the marketplace for catastrophe models and facilitate innovation by a broader knowledge group. It focuses on providing infrastructure and resources in an open, collaborative manner to scientists collecting data on various environmental conditions."
What are some of the other challenges you encounter in harnessing big data?
“One is that data quality varies significantly. Before we commit to a data source and develop tools for using it, we want to make sure the data is accurate and fit-for-purpose. Sometimes there is a discrepancy between what the data is indicating and the facts on the ground.
“In addition, we need to understand how a big data application will benefit our clients and our business. Combining accurate and specific location information with on-going weather or seismic data certainly helps us to proactively manage our response to a major event, instead of just waiting for the claims to roll in.
“Capitalizing on big data also requires some new capabilities – bringing unstructured and structured data from different sources together into a unified framework involves unique skills and new tools. For example, aggregating data from a satellite feed with location data from the United States Geological Survey with historic claims data into a format that can be manipulated and visualized in real time, that’s a new challenge for us.
“As a result, I expect we and other global insurers will bring people on board with backgrounds not ordinarily associated with the insurance industry. For example, we recently had a very successful experience with someone from an urban planning background. In addition to being very skilled in working with large data sets, he could create visualizations drawn from different data sources that added immensely to our understanding of the exposures in certain portfolios.”Last question. Do you think big data will have an incremental or transformative impact on the insurance industry?
“Transformative! The use of big data among insurers is a relatively new development, and the benefits are just starting to emerge. Nonetheless, I think those insurers that successfully deploy big data will achieve significant competitive advantages over those that don’t. Why? Because big data can enable insurers to identify and understand risk more deeply, accurately, quickly and productively.”
Want to know more? You can reach Alan at: firstname.lastname@example.org