Research On The Measure Of Non-life One-year Reserve Risk

Solvency II brings great changes to the international insurance industry as a set of standardized management system. It represents a new trend of development of international insurance regulatory affairs, and lots of countries in the world pay attention to this new solvency requirement and try to learn from that. Nowadays China is in the process of establishing the new solvency regulation system, we should thoroughly discuss the of innovation theory and specific methods of Solvency II combining with our national conditions, and learn from international advanced experience and reforming achievements, in order to perfect our country’s insurance industry regulation, and update its supervision system.This paper presents the one-year reserving risk under Solvency II framework, trying to illustrate the necessity of its supervision. We present the one-year reserve risk in the Bayesian Multivariate log-normal model. The model s a Bayesian stochastic claims reserving model that considers simultaneously claims payments and incurred losses information and allows for deriving the foil predictive distribution of the outstanding loss liabilities. In this model we study the conditional mean square error of prediction (MSEP) for the CDR uncertainty, which is the crucial uncertainty view under Solvency II. Within our framework we are not only able to calculate the conditional MSEP for the CDR but we can also derive the predictive distribution of the CDR via MC simulations. We put forward the countermeasures and suggestions which have important significance as to the accuracy and sufficiency of the reserves liability.Finally, actuarial practice data is taken in empirical analysis. And the analysis results are compared with the results of Bayesian log-normal model and CLR method. MSEP of CDR is relatively small calculated under Bayesian multivariate lognormal model, drawing a conclusion that this model can make full use of the known data information, and be more effective in measuring one-year reserve risk. Moreover, the empirical density of CDR is compared to the Gaussian density, finding that these two densities look similar but not absolutely coincident. Further studies are needed on the assumed density of CDR.