The Study And Application Of Catastrophe Model For Earthquake Insurance

Catastrophe insurance can effectively transfer and disperse devastating losses caused by earthquake disasters.While the occurrences of destructive earthquakes,or the so-called “black swan events,” are typically of low frequency,high consequence and are unpredictable in nature,the claim payment amount can hence be extremely volatile.This loss volatility can put(re)insurers at great risk.Reasonable insurance actuarial analysis and adequate pricing are the keys to ensure(re)insurance companies to conduct their catastrophe insurance business continuously even in the face of catastrophic losses.The traditional actuarial method relies primarily on the “Law of Large Numbers” to evaluate the loss of a large amount of historical loss experience.When coming to assess the seismic risk,the analysis often leads to large deviations.Hence,the analysis results are often challenged and the reliability of the analysis is greatly reduced.According to both China domestic and international earthquake insurance practice,the catastrophe model(cat model)constructed based on physical mechanism and insurance technology is generally considered to be an effective tool for seismic risk pricing and has been widely utilized by the(re)insurance industry.Since the seismic activity is characterized by high intensity,wide distribution and long recurrence period in China,it is close to impossible to directly duplicate the existing cat model developed abroad.Researching and developing a localized earthquake cat model hence becomes necessary to reflect the China's national conditions.Therefore,this dissertation constructs a cat model,EQCat,focusing on seismic risk assessment for earthquake insurance,and studies several key issues in the development of local catastrophe models.The major achievements made in this dissertation are as follows:(1)This dissertation reviews the development history of earthquake insurance in China,and summarizes the progress of earthquake cat model research and development in China and abroad.It systematically expounds the specific contents of the four components of the earthquake cat model: Hazard Module,Exposure Module,Vulnerability Module and Financial Module.In addition,the difficulties and challenges encountered currently are discussed in each part of the model.(2)An earthquake catastrophe insurance model,EQCat,with good scalability is constructed and the system design and coding implementation are completed using MATLAB.EQCat is capable of providing technical support for seismic risk pricing and insurance loss actuarial analysis.(3)A strong motion data platform for engineering application is developed.It includes two systems,one being the strong motion record database with selecting characteristic analysisand the other a system that allow users to select ground motion records for specific engineering requirements.Moreover,a new method of acceleration baseline correction that is based on L1 norm regularization is proposed.This platform can provide data for developing regional attenuation relationships and for searching the input records for seismic vulnerability analysis.(4)This dissertation proposes a new building classification system that is based onthe factors including Construction Class,Occupancy Type,Building Height(No.of floors),Year Built,Seismic Fortification Intensity,and Geographic Region.The classification takes into consideration different levels of data availability in the world of insurance underwriting and has the characteristics of being hierarchical,expansive and comprehensive.It is one of the bases for the cat model to etimate the building loss.(5)Seismicdamage data are collected for 118 destructive earthquakes in mainland China from 1990 to 2014.The characteristics of typical structure damage are analyzed for different regions in different time periods.This dissertation proposes a method to transform the seismic damage matrices with ground motion intensity into vulnerability curve with peak ground acceleration(PGA).The proposed method takes into account the uncertainty involved in converting intensity to PGA.Damage data after transformation provideprovide the vulnerability curves that can be used in EQCat for loss estimation.(6)The impact of stochastic events,attenuation relation and vulnerability curves on loss estimation is analyzed in EQCat.It is found that vulnerability imposes the biggest impact,followed by attenuation relations and stochastic events(different simulation time with 100,000 years and various 10,000 years).