Many organizations with significant exposures to natural and man-made catastrophes struggle to quickly and accurately assess the impact of those disasters on their business.

I still remember the days when a major hurricane, flood or earthquake would trigger a fire-fighting exercise for my entire analytics staff. We had to answer many questions - about the event's impact on insurers business and the supply chains they insured - and we had to answer them quickly. Estimating the financial impact of catastrophes required sophisticated tools, trained specialists, and detailed information.

Recent developments in technology and availability of software as a service (SaaS) catastrophe alerting systems, though, have helped overcome many of the hurdles to building "real-time" response tools for decision makers. Fully automated response systems can now estimate the financial impact of catastrophes in minutes.

Building blocks of a "real-time" catastrophe response system

Bringing together 3 key building blocks with automation enables insurers to generate quick insights and overcomes a challenge even for those heavily invested in this effort.

• Exposure data. Getting the right data on what is exposed to catastrophes is the most critical building block, and surprisingly also the hardest for large global re/insurers. Data should be appropriate for the analysis, detailed enough to answer all the questions that come up, and free of errors and omissions. Most importantly, it needs to remain current through frequent updates. This is especially important for insurers and reinsurers whose portfolio of risks is constantly changing due to expiration of policies and addition of newly written business.
• Catastrophe event data. Getting the most up-to-date information on catastrophes and in the right format is the next important building block. Obtaining this information has become relatively less challenging in developed countries, where government and private organizations collect and publish real-time event information. Seismic, weather and wildfire data are available from organizations such as USGS, NOAA, NWS and GeoMAC in multiple spatial formats and RSS feeds.
• Analytics Engine & Applications. This third block is a geospatial analytics engine for identifying impacted exposures by overlaying catastrophe event "footprint" on the entire portfolio of exposures. It should be able to quickly process large sets of policies and locations subject to complex insurance contract terms, and display results on an interactive map-based user interface such as Google or Bing maps. Additional analytics, reports and alerting mechanisms are built on top of this to serve an organization's decision- making needs.

One successful case study

A simplified illustration of the real-time cat response system that we recently implemented is shown below.


In this system, insurance exposure data is collected by means of a "data pipeline" which was built to pull data from a third-party policy system into our company data warehouse (on MS Azure), and from there to the data warehouse (on AWS) of the real-time response app powering our analytics as explained below. Despite the data making multiple hops through various "clouds," it is refreshed at close of business daily.

For the catastrophe event data and spatial analytics app, we licensed a subscription-based SaaS platform to integrate into our response system. This platform - EigenPrism (from EigenRisk.com) - gives us access to a multitude of historical and 24/7 real-time worldwide events covering 12 major perils, sourced from government and private organizations. Leveraging the power of high-performance computing in the cloud, the platform is able to process millions of locations in a matter of seconds. It overlays event "footprints" on our insurance portfolio, estimates the impact, and sends out alerts via emails and smart phone notifications; allows for probability and proximity analysis to anticipate potential impact of a wind storm or wildfire as it progresses; determines the losses to various participants in a re/insurance contract, and allows for what-if analysis by modifying footprints and damageability assumptions on the fly.

The output from this app feeds into critical business decisions: identifying areas for new business suspension, loss prevention services, claims management, reserving and internal/external communications.

Technology to the rescue

Over the last ten years the world has suffered USD1.8 trillion in economic losses, USD500 billion in insurance losses and nearly 800,000 fatalities from natural disasters. In an effort to mitigate these losses, there have been sustained efforts to build early warning and disaster forecasting systems. Much of this information is openly shared "online" by many countries and non-profit agencies.

The downstream application of event data feeds for risk assessment and mitigation decisions, though, has lagged behind due to data and IT limitations within organizations exposed to catastrophe risks. With the advent of powerful software-as-a-service platforms on cloud that combine spatial and big data analytics, there are hardly any technological hurdles today to pull together a real-time catastrophe alerting and loss estimation tool.

Once a comprehensive event assessment and response system in place, insurance companies (and brokers) will not only be able make better risk management decisions for themselves, but also share relevant insights with their customers in real-time. This is the next step in the evolution of these capabilities.


If you would like to hear more from Guru RAO, join the "Chief Data & Analytics Officer, Insurance" conference taking place from March 26-29 in Chicago.  Visit the event's website for more information. The conference is organized by CORINIUM Intelligence and supported by XPRIMM Publications as Media Partner.