Research On Risk Measurement For Non-Life Insurers

As an important part of financial system (especially insurance system), non-life insurers are playing a more and more important role in economic loss financing because of their professional risk management. So risk profile of non-life insurer becomes significant in its internal risk management, also is concerned by related interest counterparts, such as policyholders, supervisors or regulators, rating agencies and so on. Risk measurement for non-life insurers is focused on in order to apply several certain accurate, scientific and feasible risk measures into practice of non-life insurers'risk management.Firstly, there are four topics overviewed including principles of risk measure, risk measures, estimation and risk measures for non-life insurers. Furthermore it is concluded that risk measurement for non-life insurers should be guided by thoughts of"Enterprise Risk Management"(ERM). In a short term, risk measures should abide by both"theoretical credibility"and"practical feasibility", which means that some basic risk measures used by non-life insurers had better be guided by certain theoretical framework (e.g. ERM) and also easy to apply to risk management.It is found that Expected Shortfall (ES) should become a basic risk measure for non-life insurers as a result of its good mathematic characteristics. A new risk measure, Non-Recovery Ratio (NRR), is introduced based on ES. The coordination between ES and NRR could help non-life insurers make more creditable management decision.Thereafter risk estimators of ES and NRR are got in different estimation methods based on Monte Carlo simulation. It is concluded:1. For a univariate risk, its ES and NRR estimators can be accurately got in both parametric methods and nonparametric methods if samples are large enough; otherwise parametric methods is preferred to nonparametric under the environment of small samples before model risk can be avoided;2. For a compound risk, its ES and NRR estimators can be accurate enough in parametric methods based on approximation function, as well as nonparametric method under the environment of large samples;3. For an independent multivariate risk, its ES and NRR can be got in ways of conjunction and moment generation function. For a dependent multivariate risk, copulas can usually be used to estimation of ES and NRR, but caution of model risk.Finally results above are applied to risk measurement of Chinese non-life insurers'cash flow, and ES and NRR estimators are calculated. It is concluded long-term cash flow of Chinese non-life insurers would be worsen if expense ratio lose control, although short-term cash flow is rather safe.