[INSURANCE] Trends in extreme extratropical cyclones and non-indemnity insurance risk

Lead Research Organisation: University of Exeter
Department Name: Engineering Computer Science and Maths


The main aim of this collaborative project is to quantify the trends in extreme extratropical cyclones in Europe and the related impacts on the effective design of non-indemnity reinsurance contracts such as catastrophe bonds. The project combines ideas from atmospheric and climate science with statistical techniques and models to tackle a problem in finance and insurance in a new, exciting interdisciplinary approach. The studentship will be jointly managed by the University of Exeter and Willis Re, a leading global insurance and reinsurance broker. Climate change is expected to affect the behaviour of extra-tropical cyclones. Recent studies have used ensembles of coupled models or high resolution coupled models to look at changes in intensity. There remains, however, a large amount of uncertainty in the regional and local predictions of changes in storminess, partly due to the spatio-temporal variability (on multi-annual to multi-decadal scales) of the physical characteristics of extratropical cyclones. Moreover, many measures of storminess used in previously published scientific studies (such as band-pass filtered storm track) are not directly related to extremes in surface wind speeds, of relevance for assessment of insurance-related windstorm risk in Europe. The idea of this project is to use measures of storminess which are directly relevant for current practice in non-indemnity insurance. In non-indemnity insurance, a proxy measure of insured loss is used instead of the actual insured losses. The trigger for payment may be an index (e.g. based on wind speed), or parametric (e.g. based on earthquake magnitude), or even based on model output. Non indemnity insurance products include Industry Loss Warranties, Catastrophe Bonds, or simply an Index Product. There is growing interest in non-indemnity insurance, also because it opens up the potential investor base to the Capital Markets, where investors look for non-correlated assets. Our basic approach is to develop simplified versions of catastrophic windstorm models which allow us to study the effect of trends and spatio-temporal variability in extreme extratropical cyclones on insurance products of the above described nature. Data will be used from the ERA-40 and ACRE reanalyses and from an ensemble of runs of the Met Office global climate model (HadCM3). As a proxy for insurance exposures, we will use the LandScanTM database of worldwide population. This information will be fed into the simplified loss model, yielding modelled loss time series. These will be used to assess the robustness of various parametric trigger structures taking into account the trends in spatial and temporal variability of the storms. The student will be based mainly at the Exeter Climate Systems research centre at University of Exeter, which provides an inspiring research environment for doctoral students. He/she will receive training in the quantification of risk due meteorological hazards and in the mathematical and statistical modelling of complex weather and climate processes. The student will learn to manipulate large meteorological datasets and perform analysis using state-of-the-art statistical models. The student will also benefit from working under the supervision of the chief actuary at Willis Re (the CASE partner), gaining first-hand experience in modern insurance industry practices for modelling risk of large losses due to natural catastrophes. This combined training in academic and industry relevant skils will provide the student with a large range of employment opportunities. Benefits for the CASE partner include enhanced risk management through more appropriate quantification of basis risk in non-indemnity products, which is important in view of the capital and risk management requirements in the Solvency II Framework Directive that was adopted by the European Parliament's plenary session on 22 April 2009, with an implementation date of 31 October 2012.


10 25 50