손해보험 지급준비금 위험조정 실증분석 - 음수의 추가지급보험금이 존재하는 경우

This paper deals with a method of measuring the variability of loss reserve in non-life insurance, which can be interpreted as the risk adjustment of the liability for incurred claims and the liability for remaining coverage in the framework of K-IFRS. Although negative incremental claims appear frequently in the loss development triangles, most of the classical loss reserving estimation methods fail to provide complete solutions when there are many negative incremental values. The alternative approach proposed by Kunkler(2006) will be adopted to model the negative incremental claims, where a generalized linear model will be applied to model the sign and the severity of the incremental loss data. Numerical example of real industry data will be provided, which cannot be found in the previous literature. Bayesian approach will be adopted to estimate the proposed model and the technical details of setting up the design matrix will be discussed. Gibbs sampling and Markov chain Monte Carlo techniques will be applied in MATLAB to implement the model. As an observation of the empirical analysis, the importance of model selection will be emphasized.