考虑两类赔款数据相关性的随机性准备金进展法及改进

中国管理科学 ›› 2014, Vol. 22 ›› Issue (4): 9-16.

考虑两类赔款数据相关性的随机性准备金进展法及改进

段白鸽¹, 张连增²

1. 复旦大学经济学院, 上海 200433;
2. 南开大学经济学院, 天津 300071

收稿日期:2011-08-09 修回日期:2014-02-06 出版日期:2014-04-20 发布日期:2014-04-23
作者简介:段白鸽（1983- ），女（汉族），山西临汾人，复旦大学经济学院教师，博士，师资博士后，中国准精算师，研究方向：精算与定量风险管理、不确定经济学.
基金资助:
国家自然科学基金面上资助项目（71271121）；中央高校基本科研业务费专项资金（NKZXTD1101）

Stochastic Reserves Development Methods Based on the Correlation Among Paid-Incurred Payments Data and Their Improvement

DUAN Bai-ge¹, ZHANG Lian-zeng²

1. School of Economics, Fudan University, Shanghai 200433, China;
2. School of Economics, Nankai University, Tianjin 300071, China

Received:2011-08-09 Revised:2014-02-06 Online:2014-04-20 Published:2014-04-23

摘要/Abstract

摘要： 本文创新性地提出基于成对抽数的非参数Bootstrap方法、二元正态分布和Copulas函数三种考虑已决赔款与已报案赔款相关性的随机性准备金进展法，并结合非寿险精算实务中的经典流量三角形数据，应用R软件对三种考虑相关性的随机性准备金进展法进行了完整的编程实现，并模拟得到了最终损失、未决赔款准备金和IBNR的完整的预测分布。本文提出的考虑相关性的随机性准备金进展法不但考虑了两类赔款数据之间的相关性，而且体现了不同事故年已发生已报案未决赔款准备金进展情况之间的差异。这种处理相关性的思路和方法在多元准备金评估中具有重要的应用价值。

关键词: 随机性准备金进展法, 多元准备金评估, Copulas函数, Bootstrap方法, 二元正态分布, 预测分布

Abstract: Three stochastic reserves development methods are innovatively proposed in this paper considering the correlation between paid-incurred payments data, i.e. based on non-parametric Bootstrap method with pairwise resampling, based on bivariate normal distribution, and based on Copulas function. Combined the classic run-off triangles data in the non-life actuarial practice, a complete programming for three stochastic reserve development methods is provided based on the correlation with R software. Further, the complete predictive distributions of ultimate loss, outstanding claims reserves and IBNR are simulated. The proposed stochastic reserve development methods based on the correlation not only consider the correlation between the paid payments and the incurred payments, but also reflect the development difference of incurred and reported outstanding claims reserves in different accident years. Such ideas and methods considering correlation have important theoretical significance and practical value for multivariate reserving.

Key words: stochastic reserve development methods, multivariate reserving, Copulas function, Bootstrap method, bivariate normal distribution, predictive distribution

中图分类号:

F840.4

段白鸽, 张连增. 考虑两类赔款数据相关性的随机性准备金进展法及改进[J]. 中国管理科学, 2014, 22(4): 9-16.

DUAN Bai-ge, ZHANG Lian-zeng. Stochastic Reserves Development Methods Based on the Correlation Among Paid-Incurred Payments Data and Their Improvement[J]. Chinese Journal of Management Science, 2014, 22(4): 9-16.

参考文献

[1] England P D, Verrall R J. Analytic and bootstrap estimates of prediction errors in claims reserving[J]. Insurance: Mathematics and Economics, 1999, 25(3): 281-293.

[2] England P D, Verrall R J. Stochastic claims reserving in general insurance[J]. British Actuarial Journal, 2002, 8(3): 443-518.

[3] England P D, Verrall R J. Predictive distributions of outstanding liabilities general insurance[J]. Annuals of Actuarial Science, 2007, 1(2): 221-270.

[4] England P D. Addendum to "analytic and bootstrap estimates of prediction errors in claims reserving"[J]. Insurance: Mathematics and Economics, 2002, 31(3): 461-466.

[5] Clark D R. LDF curve fitting and stochastic loss reserving: a maximum likelihood approach[J]. Casualty Actuarial Society Forum, 2003, (3): 41-91.

[6] Meyers G. Estimating predictive distributions for loss reserve models[J]. Variance, 2007, 1(2): 248-272.

[7] Guszcza J. Hierarchical growth curve models for loss reserving[J]. Casualty Actuarial Society E-Forum, 2008, (3): 146-172.

[8] Wüthrich M V, Merz M. Stochastic Claims Reserving Methods in Insurance[M]. New York: Wiley, 2008.

[9] Peters G W, Shevchenko P V, Wüthrich M V. Model uncertainty in claims reserving within Tweedie's compound Poisson models[J]. ASTIN Bulletin, 2009, 39(1): 1-33.

[10] Björkwall S, Hössjer O, Ohlsson E, et al. A generalized linear model with smoothing effects for claims reserving[J]. Insurance: Mathematics and Economics, 2011, 49(1): 27-37.

[11] England P D, Verrall R J, Wüthrich M V. Bayesian over-dispersed Poisson model and the Bornhuetter-Ferguson claims reserving method[J]. Annals of Actuarial Science, 2012, 6(2): 258-283.

[12] 张连增. 未决赔款准备金评估的随机性模型与方法[M]. 北京:中国金融出版社, 2008.

[13] 张连增, 段白鸽. 非寿险索赔准备金评估随机性方法[M]. 北京:北京大学出版社, 2013.

[14] Liu Huijuan, Verrall R J. Bootstrap estimation of the predictive distributions of reserves using paid and incurred claims[J]. Variance, 2010, 4(2): 121-135.

[15] Happ S, Merz M, Wüthrich M V. Claims development result in the paid-incurred chain reserving method[J]. Insurance: Mathematics and Economics, 2012, 51(1): 66-72.

[16] Happ S, Wüthrich M V. Paid-incurred chain reserving method with dependence modeling[J]. ASTIN Bulletin, 2013, 43(1): 1-20.

[17] 张连增, 段白鸽. 准备金评估的随机性Munich链梯法及改进——基于Bootstrap方法的实证分析[J]. 数量经济技术经济研究, 2011, 28(11): 98-111.

[18] 张连增, 段白鸽.未决赔款准备金评估的随机性Munich链梯法[J]. 数理统计与管理, 2012, 31(5): 880-897.

[19] 张连增, 段白鸽.基于已决赔款与已报案赔款相关性的随机性准备金进展法[J]. 管理评论, 2013, 25(5): 155-166.

[20] Zhang Yanwei, Dukic V, Guszcza J. A Bayesian non-linear model for forecasting insurance loss payments[J]. Journal of Royal Statistical Society, Series A, 2012, 175(2): 637-656.

[21] Merz M, Wüthrich M V, Hashorva E. Dependence modeling in multivariate claims run-off triangles[J]. Annals of Actuarial Science, 2013, 7(1): 3-25.

[22] 吴小平. 保险公司非寿险业务准备金评估实务指南[M]. 北京:中国财政经济出版社, 2005.