Title: The study of modelling problems of GLM for outstanding claims reserving based on prediction error of reserve

Authors: Chun Yan; Yiwei Qiu; Liangyu Zhang; Wei Liu

Addresses: College of Mathematics and Systems Science, Shandong University of Science and Technology, 579 Qianwangang Road, Qingdao, China ' College of Mathematics and Systems Science, Shandong University of Science and Technology, 579 Qianwangang Road, Qingdao, China ' College of Mathematics and Systems Science, Shandong University of Science and Technology, 579 Qianwangang Road, Qingdao, China ' College of Information Science and Engineering, Shandong University of Science and Technology, 579 Qianwan gang Road, Qingdao, China

Abstract: Stochastic methods for outstanding claims reserving based on generalised linear model (GLM) framework are the hotspot of actuarial theory. Based on the GLM framework, the mean square error of prediction of reserves could be calculated. Focused on GLM for outstanding claims reserving, this paper analysed two problems. One is the distribution's assumption chosen of the response variable; the other is the over-parameterisation in models. First, a weighted function was introduced to judge the effect of the model fitted under different distributions. Second, referring to Cape-Cod method, a three-parameter GLM was built by introducing more claims information to reduce the number of parameters in the predictor structure. At last, examples were given to prove the effectiveness of these methods.

Keywords: outstanding claims reserve; generalised linear models; GLM; prediction error; stochastic methods; over-parameterisation; modelling; actuarial theory.

DOI: 10.1504/IJRIS.2015.072946

International Journal of Reasoning-based Intelligent Systems, 2015 Vol.7 No.3/4, pp.193 - 199

Received: 01 Dec 2014
Accepted: 16 Jan 2015
Published online: 09 Nov 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article