Mr. Schmidl, you head one of five pilot projects being conducted by Munich Re to find out how using big data can optimise risk assessment and pricing. Tell us more.
Our "supply chain case" analyses supply chains and aims to more accurately identify interdependencies within global industries. Manufacturers of electronics and other hi-tech companies for instance no longer possess own production sites, but have instead outsourced almost completely their production of components. While companies have gone to great efforts to make their supply chains more transparent since the 2011 earthquakes and floods in Japan and Thailand, the information available to us for assessing interdependency and contingent business interruption risks is still usually not enough to allow adequately the evaluation of supply chain disruptions into the insured´s business interruption.
Do the insureds not have this information?
In fact, much of the information that helps us during a risk analysis is not actually available to the policyholders themselves. Or other issues create hindrances, such as non-disclosure agreements or intellectual property issues. We intend to reduce underwriting uncertainty as much as possible using big data.
How do you go about assessing these risks?
We want to use also information that is already freely available on the Internet, but not structured. This includes the precise geographical location as well as details published on the web or in corporate communications concerning supply companies, their critical products and international business partners. The trick is to pick out the relevant data, scrutinise them more closely and link them with our own internal data.
Test scenarios promise to create greater transparency in business-interruption risks. What would such a scenario look like?
Our test scenario aims to make the supply chains of hi-tech electronic industry risks more transparent, such as those of smartphone manufacturers. The interdependencies in this sector are often extreme: In many cases, a single company produces one special component which it then supplies to several OEMs (original equipment manufacturers). If a supplier suffers a disruption in supply, for any reason whatsoever, this could affect production at a number of OEMs. Our aim is therefore to identify critical products and suppliers, as well as such interdependent relationships. Ideally, if we can improve our understanding of these networks, we will ultimately know how high our policyholders' business interruption risks may be.
A technical challenge.
Indeed. We also use webcrawlers, i.e. search engines programmed for specific terms, as well as knowledge models produced by our IT service provider to trawl through these vast volumes of data. In time, they optimise themselves, as the underlying terms, the metadata, become increasingly precise and comprehensive. The plausibility of the results naturally has to be validated in detail.
When are the first results expected?
The projects are scheduled to run until the second quarter of 2015, i.e. a little more than a year after the first interdisciplinary kick-off including the orientation phase. Basically, however, it is a feasibility study. First of all, we must see what can really be achieved, and which solutions could be implemented in the long term. The ratio of costs and benefits must naturally also be appropriate. If a scenario proves to be impossible or uneconomical, it is discarded and we move on to another case.
Which other industries could be analysed in this way?
Cloud services and their providers are also subject to a considerable accumulation risk, should a provider fail or be hacked. These dependencies must be made more transparent in our risk assessment. The third stream scenario refers to the oil and gas industry as well as pharmaceuticals. In such specialised industries, it is often difficult to obtain an overall picture of the main supply flows and their values. An analysis of the data available on the Internet could also be helpful here.
What added business value do you expect this to yield, if all goes according to plan?
If all four scenarios can be implemented successfully, i.e. if risk selection, capacity setting and pricing in both facultative and treaty business, as well as accumulation control in treaty business, can indeed be comprehensively improved through big-data-based underwriting, and if we can also avoid losses due to unprofitable business, we estimate that this could yield a potential equivalent to revenue in the order of several tens of millions.
How will our clients benefit from these activities?
We will share the results of our analyses with our clients. The better the data, the more accurate we can make our risk assessment – and theirs, too. This, in turn, will enhance the accuracy of our pricing. We can better understand the complex risks and offer specific solution tailored to the client's needs – a clear competitive advantage for our clients and for us.