15 // THE DISASTER GAP: HOW INSURERS AND THE CAPITAL MARKETS CAN HARNESS BIG DATA TO CLOSE THE GAP


Dickie Whitaker, director of the Lighthill Risk Network and Oasis Loss Modelling Framework sounds a note of caution in relying too much on the output of catastrophe models. “The insurance and reinsurance industry has an emerging understanding that these models are very uncertain. But I’m not convinced some of the investors in the ILS marketplace have that same appreciation.”


“For the investor base supporting the cat bonds and the ‘quants’ that do all the work, how can they get a real understanding of the uncertainty in the models when it isn’t available from the cat modelling companies today in a format we want to use?” he continues. “They see returns that don’t incorporate the full uncertainty that we’re all aware are in these products.”


One of the biggest difficulties as the cat bond market looks to grow and diversify is the lack of data and catastrophe models in other geographical regions. Many regions lack both historical hazard data and claims information upon which catastrophe underwriters would traditionally turn to. In these situations the industry must look to other sources of data to help assess and underwrite the risk.


While cat modellers, brokers and reinsurers continue to invest in modelling and analytics, it is clear a different approach is needed if cat bond capacity is to be successfully deployed to cover a broader range of territories and risk. The exploitation of big data which is loosely defined as voluminous, fast changing and includes unstructured data could be one of the keys to achieving this.


“Only by joining together multiple different structures of different datasets and using some fairly sophisticated algorithms can you really understand the risk,” says Alex Plenty, associate partner, business analytics and optimisation, Global Business Services at IBM.


“If investors are serious about taking an opportunity to differentiate or to compete they will be looking at these things in more detail,” he continues. “You can use a default model or a parametric index or you could use some other method to give yourself the confidence. But the days are gone when going by the output of a third party is enough. If you’re a sophisticated investor with serious intent in this market you need to be taking some time to understand the risks more fully.”


Forward-thinking market participants are looking to “non-traditional” data sources and processing techniques. For example, collating data from numerous models (external and internal) to facilitate sensitivity analysis and comparison.


Catastrophe models typically use a stochastic model for estimating probability distributions of potential outcomes by allowing for random variation in one or more inputs over time. The random variation is usually based on fluctuations observed in historical data for a selected period using standard time-series techniques. Distributions of potential outcomes are derived from a large number of simulations (stochastic projections) which reflect the random variation in the input(s).


Big data facilitates the analysis of actual weather readings allowing the construction of predictive models. The combination of legacy model results with predictive models produces more robust frequency and severity attributes for a set of events relating to particular perils. Other developments include:


– Looking at open and transparent frameworks to understand uncertainty (eg Oasis Loss Modelling Framework);


– Augmenting data with highly unstructured and volatile information;


– Improving speed of analysis by the use of Massively Parallel Processing, complex in-database algorithmic analysis; and


– Intuitive visualisation such as Geospatial and dashboarding.


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24