Aon Benfield Research Newsletter

Aon Benfield Research Newsletter A message from our Aon Benfield Research Steering Committee Chair Non-modelled perils have driven a plethora of research over the last twelve months to boost the insurance industry’s understanding of these hazards. This theme will very much continue throughout 2013 as, where possible, we seek to support the transition of such perils into the ‘modelled’ domain. We also plan to broaden our research interests into areas of uncertainty in modelling by being part of the PURE research initiative and through working with new partners to explore possible collaborations between business, academia and international nongovernmental organisations. With the third annual Aon Benfield Research Conference taking place in Singapore this year — one of the world’s most influential insurance and reinsurance hubs — there will also be a shift in focus towards Asia-Pacific. Our goal is to further engage with our partners to share global expertise and tackle the specific hazard risks in the rapidly-developing region. In this edition of the newsletter, we introduce our new partners Ambiental, Deltares and Oscilmet, look at how the latest academic research is being incorporated into catastrophe models for North Africa earthquake and Switzerland flood, and interview Gottfried Grünthal from GFZ Potsdam. Enjoy!

Catherine Tillyard Chair of Aon Benfield Research’s steering committee [email protected]

Edition 4 — 2013

Table of Contents 2 Enhancing catastrophe models through research

New hailstorm tracking tool

3 Estimating maximum earthquake magnitude Downscaling European windstorm data Tackling uncertainty in models 4 2nd Aon Benfield Research Conference

New members

5 Member spotlight: Aon Benfield UCL Hazard Centre 6 Can we trust earthquake cluster detection tests? 8 Interview with Gottfried Grünthal

Aon Benfield Research Newsletter

Edition 4 — 2013

Innovation Enhancing catastrophe models through research Over half of Aon Benfield Research partners work with Impact Forecasting, Aon Benfield’s model development centre of excellence, to create model components. Switzerland flood New partner Ambiental, a leading international flood risk assessment consultancy, was instrumental in developing Impact Forecasting’s flood model for Switzerland and Lichtenstein to quantify financial losses caused by riverine and lake flooding. Flooding in Switzerland over the last two decades has caused human and financial losses across the entire country, for example in 1999, 2000, 2005 and 2007. As such, flood is considered to be the most significant natural peril in Switzerland. Ambiental’s Flowroute-iTM, a 2-dimensional hydrodynamic flood model, was used to improve the speed and efficiency of digital flood map creation using advanced computing and data processing capabilities. Petr Puncochar, flood model developer at Impact Forecasting, said: “The hazard component of the Swiss flood model is based on advanced hydrodynamic modelling, which describes the real physical behaviour of floods with improved accuracy. In particular, this approach enables better understanding of the complex Swiss topography which includes various hydrological regimes from low land to alpine.”

Depth [m] Max: 10.00 Min: 00.01 Buildings

A 1 in 100 year flood extent map for the River Ticino joining to Lago Maggiore, Switzerland (Source: Impact Forecasting)

Algeria and Morocco earthquake Equally, research from the Aon Benfield Natural Hazard Centre in South Africa was critical for the hazard component of the new Algeria and Morocco earthquake models. Local expertise was used to generate a stochastic set of 55,000 events over a 50,000 year period.

2

Since the May 2003 earthquake in Algeria, which killed more than 2,000 people, left 150,000 homeless and caused economic losses of an estimated 10% of the country’s GDP, legislation has enforced mandatory insurance against natural catastrophes for all household, commercial and industrial property risks. Following the 2004 magnitude 6.5 Al Hoceima earthquake in Morocco that killed more than 620 people, injured 900 and caused economic losses estimated at USD400 million, market discussions have arisen about the creation of an obligatory coverage for natural catastrophes. Goran Trendafiloski, earthquake model developer at Impact Forecasting, said: “As a major insured peril, it is crucial that modelling of earthquakes continues to evolve to mirror the latest science and changing insurance industry needs. The Algeria and Morocco model launch reflects Impact Forecasting’s objective to deliver cross-country models, while using locally sourced information for various model components and a better knowledge of building properties. In addition, across all territories, the team is also exploring model development for secondary perils such as liquefaction, fire following and tsunami.”

ELEMENTS: enhancing customisation of catastrophe risk assessment In addition, Impact Forecasting has launched its latest version of ELEMENTS, its loss calculation platform to enable insurers to more easily customise catastrophe models. Insurers can now incorporate their own views on the risk of property damage through customised vulnerability curves, based on their actual loss experience or current portfolio. This in turn produces estimates of potential financial losses that help to better inform and influence reinsurance buying strategies. The platform gives insurers access to Impact Forecasting’s 30 catastrophe models spanning over 20 territories and six key perils. ELEMENTS gives insight on territories worldwide, including for the peak risk zones and the newly released models for Turkish and Moroccan earthquake and German storm surge.

Scenario models Impact Forecasting has also launched a suite of new scenario models to generate loss estimates for specific historic or hypothetical events. Events such as the storm surge by Superstorm Sandy, the 2011 Thailand floods, and the highest insured loss European windstorm, Kyrill, can now be analysed to gauge the financial impact of their potential reoccurrence. The scenarios are generated by integrating footprints from either Impact Forecasting, insurers and reinsurers or third party organisations including Aon Benfield Research partners. Contact: Adam Podlaha ([email protected]) on ELEMENTS. Aon Benfield Research


Edition 4 — 2013

Latest Research Downscaling European windstorm data A new paper from University of Cologne will play an integral part of the hazard component within Impact Forecasting new European windstorm model due in 2013. ‘A combined statistical and dynamical approach for downscaling large-scale footprints of European windstorms’ was recently published in Geophysical Research Letters and chosen as an “Editor’s Highlight” (doi:10.1029/2012GL054014). For detailed and reliable regional impact studies, large datasets of high-resolution wind fields are required. A statistical downscaling approach in combination with dynamical downscaling is introduced to derive storm related gust speeds on a highresolution grid over Europe. The method is computationally inexpensive and reproduces observed windstorm events adequately. The technique can be applied to ensembles of global climate models to project the potential future regional impacts of wind in a changing climate. In addition, University of Cologne has published a paper with colleagues from the University of Reading in Nature Geoscience (doi:10.1038/ngeo1438) entitled ‘Response of the North Atlantic storm track to climate change shaped by ocean-atmosphere coupling’. Contact: Dr Joaquim G. Pinto ([email protected])

New hailstorm tracking tool New partner Oscilmet has launched a new hailstorm tracking service in Australia to help insurers obtain a view of the impact of an event within hours. The system called Hailmap sends immediate alerts by email and text to Aon Benfield colleagues to help insurers’ assess potential damage in relation to their exposure. This is achieved by creating an event footprint, using ImpactOnDemand — Aon Benfield’s mapping tool that allows insurers to visualise and quantify their exposure to risk.

Aon Benfield worked with Oscilmet to turn the concept into a product and develop a process where an alert is only sent once a certain number of properties could potentially be impacted. As hail falls in unpopulated places it was important to filter out that data to avoid unnecessary alerts. This tool can be applied to other countries where there is live unrestricted flow of radar data. Contact: Mike Fewings (michael.fewings@ oscilmet.com.au) or Peter Cheesman ([email protected])

Tackling uncertainty in models Aon Benfield Analytics has joined forces with a consortium of academic institutions, led by University College London (UCL) and Bristol University, plus industry partners, including Arup, Axis, Catlin and Hiscox to better understand Probability, Uncertainty and Risk in the Environment (PURE). PURE is a Natural Environment Research Council (NERC) funded programme, over four years and with GBP4 million funding. The two high-level goals are to develop new methods to improve the assessment and quantification of uncertainty and risk in natural hazards and to provide good practice guidance and standardisation across the natural hazards community. Case studies include flood risk estimation, seismic hazard analysis, tsunami models and data limitations, European extreme weather

forecast products and risk models and volcanic ash dispersion. Aon Benfield Research partners ABUHC, EuroTempest, HR Wallingford and Lighthill Risk Network are also involved. Paul Kaye, head of the international actuarial team at Aon Benfield Analytics, said: “Uncertainty, its communication and the lack of formal recognition in decision making represents one of the greatest challenges of our industry. Our involvement with the PURE initiative will support our detailed assessment of catastrophe uncertainty using our ELEMENTS platform as well as engaging with the development of broader best practice principles in this area.” Contact: Catherine Tillyard ([email protected])

Estimating maximum earthquake magnitude The Aon Benfield Natural Hazard Centre in Africa is proposing a statistically justifiable alternative approach to the currently used Bayesian procedure (Cornell, 1994) for the estimation of the maximum possible regional earthquake magnitude (mmax). The application of the current Bayesian formalism results in the likelihood function reaching its peak at the maximum observed earthquake magnitude rather than

Aon Benfield Research

maximum possible magnitude. This may lead towards underestimation of 0.5 units. Therefore, a simple ad hoc solution is proposed to estimate potential magnitudes, which has been used as a standard in the nuclear industry to estimate the probabilistic and deterministic seismic hazards for nuclear power plants. This can also be applied for developing earthquake

catastrophe models in particular when dealing with regions with less well distinguished tectonic characteristics and incomplete earthquake catalogues. Published in Research in Geophysics Vol 2. No 1 (2012) p. 46-51. Contact: Andrzej Kijko ([email protected])

3


Edition 4 — 2013

Member News 2nd Aon Benfield Research Conference Understanding risk is core to our business and Aon gives our clients access to the world’s leading minds of the natural hazards world through the Aon Benfield Research collaboration. The 2nd annual conference brought together our academic and industry partners from around the globe with expertise in a host of perils from earthquakes, tsunamis and volcanoes to flood, engineering and windstorm. The conference aimed to help academics understand what we need from them to help our clients, to share their latest research and identify future research projects that are relevant to the insurance industry. Key issues discussed included: • Creating a Thai flood catastrophe model • Building our understanding of Japanese earthquake

New members Ambiental Ambiental, a flood risk assessment consultancy, will help improve the mapping of flood extents using their advanced hydrodynamic modeling capability. The Ambiental approach describes the real physical behaviour of flood phenomena with improved accuracy, thus enhancing probabilistic catastrophe models and construction of historical event footprints. Dr. Justin Butler, Managing Director at Ambiental, commented: “It is always rewarding to see how the results of our flood knowledge and technology can be used practically to benefit the insurance industry, especially considering the scale of losses from recent extreme flood events internationally. Our hydrodynamic modeling tools use the latest computational techniques to increase processing speed whilst maintaining accuracy and stability so as to meet the demands of insurers, brokers and reinsurers for high resolution flood mapping.”

Oscilmet Oscilmet, provider of predictive and analytical weather services, enables insurers and other users to view the development of damaging hailstorms both during a storm and retrospectively, via an animated storm footprint — graded in intensity — which may be overlaid onto a client’s GIS.

Oscilmet’s products for the insurance market include its sensitive predictive tool, HailWarning. Already in use within the sector, HailWarning has proven its reliability by consistently predicting the path of damaging hail, enabling individual policyholders within the specified impact boundaries to be advised to protect cars and breakables.

Deltares Deltares, an independent institute for applied research in the field of water, subsurface and infrastructure, is working with Aon Benfield to develop global flood solutions. The collaboration will assist the hydrodynamic modelling of flood extents generated by Impact Forecasting. The team also licenses hydrodynamic modelling software Delft3D and there has already been a successful cooperation on tsunami for Japan. Based in Delft and Utrecht in the Netherlands, Deltares employs over 800 people who work on global smart solutions, innovations and applications for people, environment and society. Its main focus is on deltas, coastal regions and river basins. Managing these densely populated and vulnerable areas is complex, which is why they work closely with governments, businesses, other research institutes and universities at home and abroad.

• Understanding clustering of windstorms • How academic partners can support Aon’s own sustainability Keynote speaker Dominic Christian, co-CEO of Aon Benfield, reiterated the importance of research as the insurance industry expands into emerging markets such as Asia-Pacific.

Comparison of the tsunami model and measurements by Deltares: Observed (hatched) and Delft3D simulated (colour-coded) flooding levels at Talcahuano caused by the 2010 Chilean tsunami.

4



Edition 4 — 2013

Member spotlight: Aon Benfield UCL Hazard Centre Philippine volcano Mayon with remains in the foreground of one of its deadly debris flows from 2006. These flows were initiated by heavy rainfall associated with Supertyphoon Reming and they killed around 1300 people. Photo: M. Duncan 2012.

Volcanology showcase in Mexico The Aon Benfield UCL Hazard Centre (ABUHC) recently visited Mexico in order to consider volcanic risk around Mexico City and showcase its volcanology expertise at the latest Cities on Volcanoes (COV) conference in Colima. This international assembly is held once every two years in an urban district within sight of an active volcano and has the aim of utilising new science to enhance practical methods for reducing the risk from volcanic eruptions. The ABUHC made nine presentations, from forecasting eruptions and communicating hazard to evaluating community perceptions of hazard and designing new methods for managing emergencies. As a result, the ABUHC has been invited by the COV Organising Committee to co-ordinate contributions from the insurance and reinsurance, humanitarian and development sectors for the next assembly in Yogyakarta, Indonesia in 2014. This will be an ideal opportunity for the insurance and reinsurance community to engage directly with practitioners in volcano science, as well as civil protection agencies and humanitarian and development organisations, in order to improve risk transfer and reduction methods for the multi-hazardous nonmodelled peril of volcanism. Contact: Christopher Kilburn ([email protected]).

Multi-hazard risk assessment and reduction Collaborative research between the ABUHC and the Catholic Agency for Overseas Development (CAFOD) has identified that humanitarian and development non-government organisations (NGOs) are largely failing to incorporate multi-hazards in their policy and programmatic work, which is an issue that extends well beyond just these organisations. The project has found that NGO methodologies can account for more than one hazard acting independently, but they do not readily account for interactions between hazardous processes


To address this shortfall the research is adopting the event tree methodology to consider links between hazards, triggers of secondary hazards, generation of sequential chains of hazardous events, changing probabilities of future hazards and amplification of disasters. The method is to be tested in the Philippines by examining mud and debris flow hazards that may be triggered by volcanic, seismic or rainfall events on the slopes of Mayon volcano. The findings could have application in insurance and reinsurance for quantifying the probable links between hazardous events and their timings. Contact: Melanie Duncan ([email protected]).

Improving tsunami risk reduction and disaster response Over forty scientists, modellers, engineers and disaster planners, practitioners and policy-makers attended a workshop at University College London, jointly hosted by the ABUHC and the British Geological Survey (BGS), on Tsunami Disasters: How Effective has Science been for Mitigation, Planning and Disaster Relief? A presentation from Aon Benfield’s Impact Forecasting team on its Japan earthquake and tsunami model stimulated interest from the humanitarian and development agencies represented. This interest endorses the huge potential offered by insurance and reinsurance risk models and mapping tools to enhancing risk assessment, anticipation, preparedness and response in lesser-developed countries where these models and tools currently have limited penetration. Contact: Dave Tappin ([email protected]).

Bringing together academia, NGOs and the insurance industry 2012 saw the ABUHC continue its pioneering work with NGOs to assist them make greater use of research, hazard and risk science and insurance tools in their anticipation and mitigation of risk associated with natural hazards and exploitation of natural resources. The ABUHC organised a workshop in February 2013 on NGO — Academic — Business Collaborations for Enhancing Disaster Risk Reduction, Response and Relief in partnership with Aon Benfield, HelpAge International and CAFOD. The aim was to generate several mutually beneficial collaborative projects from this unique cross section of partners. Stephen Edwards of the ABUHC commented: “The Centre is leading the way on this type of collaboration and is recognised for its innovative research and knowledge exchange partnerships. It is working closely with Aon Benfield and NGOs to ensure that together we can respond rapidly to opportunities that should arise from future UK government and other initiatives requiring multi-sector collaboration on disaster risk transfer and reduction.” In recognition of his work to build research, knowledge and business partnerships with NGOs, particularly through the application of science, Stephen Edwards recently became one of the first academic members of the BOND Disaster and Risk Reduction Working Group (BOND is the UK membership body for NGOs working in international development). Contact: Stephen Edwards ([email protected]).

5


Edition 4 — 2013

Feature

Can we trust earthquake cluster detection tests? or a cyclical behaviour induced by a common cause. Whether or not these assertions hold true in the face of rigorous scrutiny is a different question.

Felipe Dimer de Oliveira The large-magnitude earthquakes that have occurred since 2004 (Chile 2010 and Japan 2011) come after a 40-year period of relative quiescence for events greater than Mw=8.4. The global historical time series of large earthquakes in Figure 1 reveals how striking a feature this 40-year gap is — more so considering the poor record prior 1920. Figure 1 may suggest that these large events occur in clusters lasting about a decade or so followed by longer intervals of quiescence. This would imply either that one event of this type triggers other events

Bufe and Perkins (2005) investigated this question in their article “Evidence for a global seismic moment release sequence”, in which they conclude that the observed clusters are too rare to be a mere coincidence. Amongst other results, they report a 0.5% chance of a ~40 year gap if the same number of large magnitude events were distributed at random over a 100 years interval. Six years and three mega events later, a different analysis by USGS researcher A. Michael (2011) favoured the opposite conclusion, presented in the article “Random variability explains apparent global clustering of large earthquakes”. Shearer and Stark (2012) followed concluding that “Global risk of big earthquakes has not recently increased”. Both of these studies employed a series of statistical tests to measure the similarity between the observed earthquake catalogue and what we expect from a truly random process. In a statistical test, the p-value measures how extreme a dataset is according to a pre-established hypothesis. Convention has it that a hypothesis should be rejected for

Figure 1 10

Magnitude

Mw > 8.8 9

8

7 1900

1920

1940

1960

1980

2000

p-values smaller than 0.05; larger values may rule out definitive conclusions. This criterion could be formally justified if, for example, we quantify the costs associated with a decision based on a test’s outcome. Such an approach has little to say regarding the truth of a scientific statement. There are two important concepts to understand here: variability of a random process and repeatability of an experiment. Random processes will produce samples with some degree of variation and sometimes produce extreme outcomes. If we can measure a large sample of outcomes from a process we can quantify the frequency of its extremes. Earthquakes are certainly not repeatable experiments, and the inferred inter-cluster times have a long duration compared with our historical record of 100 years, so that only two or three clusters are observed. Can we trust statistical tests applied to such scarce amounts of data? In a research paper that Paul Somerville and I wrote for the Lloyds of London, which repeated Bufe and Perkins (2005) analysis with the 2010-Chile and 2011-Japan earthquakes, we concluded that no definitive answer was possible — although there are some extreme features in the catalogue. This seemed to contradict Michael’s and Shearer and Stark’s articles, and has been the source of much discussion among seismologists. Perhaps it is necessary to admit the impossibility of an unequivocal answer with the knowledge we have at hand – scarce data and little knowledge of earth’s interior workings. We should not, however, take a defeatist attitude of admitting failure completely. In a paper recently accepted in the Geophysical Review Letters (Dimer de Oliveira (2012)), we quantified the power of the statistical tests used in the literature. The power of a test is defined as the probability of reaching a correct

Year 6



To answer this question we constructed a random process that is clustered by design, and applied the same statistical tests as in the articles mentioned above to many subsamples equivalent to 100 years of observations. The p-values obtained were used to derive the distribution that informs us with which frequency the clustered nature of this process is correctly detected. For this study, we used the standard criterion of rejecting the hypothesis if p-values were lower than 5% (0.05). In Figure 2 we show the result for a test under the hypothesis that the time between events is distributed exponentially — as expected for a Poisson distribution. Inspecting Figure 2, we can see how this test correctly rejects the hypothesis in 40% of cases; in the remaining 60% we would be lead to inconclusiveness or to wrongly accepting a hypothesis that is, by construction, false. It is important to stress that we can only conclude this because we have constructed a random number generator from which we can draw samples at will. The result shown in Figure 2 is for a particular test. There are better tests, especially the multinomial test used by Shearer and Stark that measures the timedistribution for an arbitrary number of events, which we have assessed to have a power of about 80%. Consistent with our conclusions, Shearer and Stark obtained smaller p-values for this test, but using a slightly different set of parameters. Evidently the results we describe above are underpinned by the variability of the process we contrived, given that some samples are more clustered than others. In essence, this analysis is similar to what is


Figure 2 0.5

0.4

Probability Probability

conclusion — rejecting a null hypothesis when it is false or vice versa. To do this we ask the question “if clusters existed, how frequently would these tests fail to lead to the correct conclusion?”

Edition 4 — 2013

0.3

0.2

0.1

0.0

0.0

0.2

0.4

0.6

0.8

1.0

P-Value p-value being sought by seismologists and scientists: trying to infer the nature of a process from testing a small sample. In our GRL publication, we argue that strong conclusions such as “random variability explains earthquake clusters” are overstated and don’t emphasise the limitations of the data available. We have not sought to provide unequivocal answers, as we think they may not be reached with present knowledge, but tried to introduce some objectivity on the limitations of statistical tests. We adopted the Socratic attitude of “…I know one thing, and that is that I know nothing”.

References Dimer de Oliveira, F, 2012 (in Press), Geophysical Research Letters Bufe, C. G. and D. M. Perkins (2005), Evidence for a global seismic moment release sequence, Bull. Seismol. Soc. Am., 95, 833-843 Michael, A. (2011), Random variability explains apparent global clustering of large earthquakes, Geo. Res. Lett., 38, L21301. Shearer, P. M. and Stark P. B. (2011), Global risk of big earthquakes has not recently increased, PNAS, published ahead of print December 19, 2011, doi:10.1073/pnas.1118525109

Felipe Dimer de Oliveira works as Principal Risk scientist at Risk Frontiers. He has a PhD in physics from Auckland University in New Zealand and his main interests are the study of stochastic process applied to natural catastrophe modeling and statistical inference. Contact: Felipe Dimer de Oliveira ([email protected])

7


Edition 4 — 2013

Interview

Interview with Gottfried Grünthal Gottfried Grünthal is professor and researcher at GFZ Potsdam, the national research centre for Earth Sciences in Germany. He has worked with the pioneers of engineering seismology and created an earthquake catalogue used for Impact Forecasting’s European catastrophe models. What’s your specialist area? Engineering seismology which is seismological research with a strong link to engineering application. Currently I am working on probabilistic seismic and tsunami hazard assessments, generation of regional harmonised long-term earthquake catalogues and macroseismology. How did you come to specialise in this field? It was just after the strong Friuli earthquake in May 1976, which was largely felt in Germany and even up to the shore of the Baltic Sea. This experience lead to the decision to develop engineering seismology as a field of research at the institute in Potsdam. As leader, I worked alongside W. Sponheuer who was one of the doyens of engineering seismology in Germany in the 1950s and 1960s. He succeeded A. Sieberg, one of the pioneers in the field.

How will your work benefit the insurance industry?

Where are you based and what makes this a great city?

According to feedback from the industry itself, our research helps to improve all geoscientific elements of earthquake models. This includes basic input as well as methodological aspects.

I am based in Potsdam which directly borders Berlin. Potsdam is a beautiful town with marvellous palaces and parks, surrounded by many lakes. The institute where I work has about 1000 employees and is situated in a spacious campus which dates back to 1874. The campus is on a hill surrounded by woodland. The research on this campus comprises a broad variety of geoscientific disciplines which offers great opportunities and which makes it unique.

What’s the most exciting future development in your field which could be of interest to the insurance industry? I believe that new insurance products might result from long-term timedependent probabilistic hazard assessments and new realistic probabilistic tsunami hazard assessments.

Contact: Gottfried Grünthal ([email protected])

What’s been your most memorable field trip? Hmm this is not so easy to answer. I would not call field trips to towns which badly suffered from earthquakes as most memorable. Great natural spectacles, which I have had the chance to watch, have been volcanic eruptions in Central America, Etna in Europe and on the north island of New Zealand.

Tell us about your current projects Our current projects are directly connected with our recently published EuropeanMediterranean earthquake catalogue (EMEC), where we provide a corresponding detailed analysis of magnitude dependent data completeness. We also use EMEC for probabilistic seismic hazard assessments for regions in Europe and the Near East.

8



Edition 4 — 2013

Aon Benfield Research’s Academic and Industry Partners • Ambiental — Ambiental is a leading flood risk assessment, surface water drainage and risk modelling consultancy that operates internationally and produces high detail flood risk data, flood hazard maps, perils risk data, flood modelling software and reports for insurance companies, the land and property sector, engineers, utilities and city council planning departments. • University College London: Aon Benfield UCL Hazard Centre — Europe’s leading multidisciplinary academic hazard research centre, comprising geological hazards, meteorological hazards & seasonal forecasting, and disaster studies & management.

• Deltares — An independent, institute for applied research in the field of water, subsurface and infrastructure. Throughout the world, we work on smart solutions, innovations and applications for people, environment and society. • EuroTempest — Transforms weather forecasts and observations into the specific information required to make successful live risk management decisions. • GFZ Potsdam — National research center for Earth Sciences in Germany, investigating global geological, physical, chemical and biological processes which occur at the earth‘s surface and in its interior.

• Pretoria: Aon Benfield Hazard Centre — One of the leading research universities in South Africa, it serves as a harbour of information for the engineering, disaster management and insurance industries.

• HR Wallingford — world-leading analysis, advice and support in engineering and environmental hydraulics, and in the management of water and the water environment.

• Academy of Disaster Reduction and Emergency Management — Based at Beijing Normal University, the Academy strives to be a national research base and talent center for disasters reduction, risk governance and emergency management in China.

• Lighthill Risk Network — A not-for-profit organisation, which brings together scientific research worldwide, industry (initially Insurance), government and third party organisations in exchanging risk-related expertise.

• CERDA — Based at the Technical University of Denmark (DTU), the Civil Engineering Risk and Decision Analysis research group focuses on natural hazard risks, impacts of climate change, engineering risks, life safety, sustainability and decision support.

• Matrisk — a leading consulting company specializing in risk assessment and decision support for the insurance, civil engineering and offshore industries worldwide.

• ClimateWise — a global collaboration of leading insurers, facilitated by the University of Cambridge Programme for Sustainability Leadership, focused on reducing the risks of climate change.

• Oscilmet — Oscilmet is an innovative, Australian-based IT company, specialising in the collation, analysis and interpretation of weather data. The company’s range of expertise includes predicting the probable impact boundaries of severe weather events and presenting this information real-time, in user-friendly formats. • Risk Frontiers — World leader in quantitative natural hazards risk assessment and risk management supported by the Australian insurance community. • Shanghai Typhoon Institute — Part of the China Meteorological Administration, its mission is to undertake basic and applied research related to tropical cyclones and is a base for training high-level professional personnel specialized in tropical cyclone science. • Spurr Consulting — Leading global consultancy on earthquake. • Tropical Storm Risk — Unrivalled accuracy in real time mapping and prediction of tropical cyclones worldwide. • University of Cologne — Leading research centre with a long record of wind storms diagnostics, extreme weather modeling and impact assessment.

• NTU Singapore — Institute of Catastrophe Risk Management — first multi-disciplinary catastrophe risk management research institute of its kind in Asia.

For more information, visit: http://www.aon.com/reinsurance/analytics/academic_research.jsp


9


Edition 4 — 2013

Key Contacts Catherine Tillyard Chair of Aon Benfield Research t: +44 (0)207 522 3821 e: [email protected] Americas Greg Heerde Aon Benfield t: +1.312.381.5364 e: [email protected] EMEA John Moore Head of International Analytics t: +44 20 7522 3973 e: [email protected] APAC George Attard Head of Analytics for Asia Pacific t: + 6562398739 e: [email protected]

10


About Aon Benfield Aon Benfield, a division of Aon plc, is the world’s leading reinsurance intermediary and full-service capital advisor. We empower our clients to better understand, manage and transfer risk through innovative solutions and personalized access to all forms of global reinsurance capital across treaty, facultative and capital markets. As a trusted advocate, we deliver local reach to the world’s markets, an unparalleled investment in innovative analytics, including catastrophe management, actuarial and rating agency advisory. Through our professionals’ expertise and experience, we advise clients in making optimal capital choices that will empower results and improve operational effectiveness for their business. With more than 80 offices in 50 countries, our worldwide client base has access to the broadest portfolio of integrated capital solutions and services. To learn how Aon Benfield helps empower results, please visit aonbenfield.com.

© Aon UK Limited trading as Aon Benfield (for itself and on behalf of each subsidiary company of Aon Plc) (“Aon Benfield”) reserves all rights to the content of this report or document (“Report”). This Report is for distribution to Aon Benfield and the organisation to which it was originally delivered by Aon Benfield only (the “Recipient”). Copies may be made by that organisation for its own internal purposes but this Report may not be distributed in whole or in part to any third party without both (i) the prior written consent of Aon Benfield and (ii) the third party having first signed a “recipient of report” letter in a form acceptable to Aon Benfield. This Report is provided as a courtesy to the recipient and for general information and marketing purposes only. The Report should not be construed as giving opinions, assessment of risks or advice of any kind (including but not limited to actuarial, re/insurance, tax, regulatory or legal advice). The content of this Report is made available without warranty of any kind and without any other assurance whatsoever as to its completeness or accuracy. Aon Benfield does not accept any liability to any Recipient or third party as a result of any reliance placed by such party on this Report. Any decision to rely on the contents of this Report is entirely the responsibility of the Recipient. The Recipient acknowledges that this Report does not replace the need for the Recipient to undertake its own assessment or seek independent and/or specialist risk assessment and/or other relevant advice. The contents of this Report are based on publically available information and/or third party sources (the “Data”) in respect of which Aon Benfield has no control and such information has not been verified by Aon Benfield. This Data may have been subjected to mathematical and/or empirical analysis and modelling in producing the Report. The Recipient acknowledges that any form of mathematical and/or empirical analysis and modelling (including that used in the preparation of this Report) may produce results which differ from actual events or losses. © Aon plc, 2013. All rights reserved.

Aon Benfield 55 Bishopsgate London EC2N 38D t: +44 (0)20 7088 0044 f: +44 (0)20 7575 7001 aonbenfield.com