臺灣博碩士論文加值系統

　保險費率之釐定一向是保險經營中最重要也最專業性者，費率訂定之公平與否不僅直接影響保險業者之盈虧與被保險人之權益，間接也會左右整個保險市場之經營。
有關保險費率之釐定，必須先將風險加以分類並讓具有同質性風險之被保險人負擔相同之保費，而現行分類費率之估算模式眾多，尋找ㄧ個合適且合理之估算模式應為每ㄧ位精算人員所需思考的問題，在眾多的估算模型之下多數的保險公司皆以統計方法來衡量適合於公司的估算方式，本研究以2001-2003年整體業界之統計資料嘗試以Bailey(1963)與Brown(1988)所提出之六種估算模型並結合最大準確可信度之概念來尋求最適當之估算模式。同時為與以往的文獻做比較，本研究選用從人因素中之性別、婚姻狀況、年齡作為研究之對象以探討其差異性。
而實證結果顯示六種估算模式中Bailey之加型模式與乘型模式於Buhlmann-Straub模式中的確顯著降低了其誤差值，在危險因素方面，性別與年齡間對於損失有顯著之差異性，婚姻狀況對於損失則無顯著差異，另外於六種估算模式中結合最大準確可信度理論，的確顯著降低了各類別間係數之差異，以及具有衡量危險分類效率化之功能，同時於Buhlmann-Straub模式下所釐定之費率係數因同時考量加權後之危險單位數與資料年度數，因此其估計之結果將更精確客觀，對於社會大眾將更具說服力。

Insurance ratemaking is usually the most important and professional in insurance management. Whether the rate is fair or not directly affects the profits of insurance industry and rights of insured and indirectly influences the management of all the insurance markets.
With regard to making insurance rating, risks should be first classified and the insured with homogenous risks should pay the same premium. Nowadays, there are a variety of mathematical methods for estimation of rating classification. Therefore, to search for a suitable and reasonable estimation model is the question each acturial has to think of. Most insurance companies use the most suitable model based on statistic methods. This paper intends to find the most suitable model that suggested by incorporating the six frequently used models with credibility of the greatest accuracy according to 2001-2003 statistic data . In comparison with the previous references, this research is centered on a person’s gender, marital status, and age in order to get an understanding of the differences of each danger factor and the appropriateness of each model.
The experiments prove that among the six models the additive and multiplicative models of Bailey indeed reduce the error. As to the danger factor, there are manifest differences of loss in gender and age while there is no difference of loss in marital status. Besides, the theory of combining the greatest accuracy credibility among the six estimation models really reduces the differences of each relativity and has the function of efficiency of measuring danger classification. What’s more, the result of estimation under Buhlmann-Straub model will be more accurate, objective and convincing, for its relativities of data consider the danger units and the data years.

Keywords：Classification Rating , Additive Model , Multiplicative Model , Greatest Accuracy Credibility , Rate Relativity

目 錄
　　　　　　　　　　　　　　　　　　　　　　
中文摘要 ------------------------------------------------------------------------------------- i
英文摘要 ------------------------------------------------------------------------------------- ii
誌謝 ------------------------------------------------------------------------------------- iii
目錄 ------------------------------------------------------------------------------------- iv
表目錄 ------------------------------------------------------------------------------------- vi
第一章、緒論 ---------------------------------------------------------------------------- 1
第一節 研究動機與目的-------------------------------------------------------------- 1
第二節 研究架構----------------------------------------------------------------------- 2
第三節 研究限制----------------------------------------------------------------------- 3
第四節 研究內容----------------------------------------------------------------------- 3
第二章、危險分類與文獻回顧 ----------------------------------------------------------- 5
　　第一節 危險分類因素之考慮要件-------------------------------------------------- 5
　　第二節 危險分類因素之分析--------------------------------------------------------- 6
第三節 文獻回顧----------------------------------------------------------------------- 9
第三章、從人因素費率釐定及探討------------------------------------------------------ 17　
　　第一節 問題探討 --------------------------------------------------------------------- 17
　　第二節 損失率法與純保費法------------------------------------------------------- 17
第三節 乘型與加型模式之概述---------------------------------------------------- 19
第四節 分類費率估算模式之探討------------------------------------------------- 20
第五節 模式之檢定-------------------------------------------------------------------- 29
第六節 回復平衡---------------------------------------------------------------------- 30
第四章、可信度理論之探討 -------------------------------------------------------------- 34
　　第一節 可信度因素值之概述-------------------------------------------------------- 34
　　第二節 有限變動可信度-------------------------------------------------------------- 35
第三節 最大準確可信度-------------------------------------------------------------- 38
第五章、資料分析與實證結果之評估--------------------------------------------------- 43
　　第一節 資料之分析與調整----------------------------------------------------------- 43
　　第二節 估計結果分析----------------------------------------------------------------- 47
第六章、結論與建議----------------------------------------------------------------------- 72
第一節 研究過程探討----------------------------------------------------------------- 72
第二節 結論與建議------------------------------------------------------------------- 74
參考文獻 ------------------------------------------------------------------------------------- 79
附錄一 ------------------------------------------------------------------------------------------ 82

參考文獻
1.林昆余，2001，第三人責任險純保費可信度估計與貝氏估計之比較，逢甲大學，統計與精算研究所碩士論文。
2.林進田等著，2000，高等產險精算理論與實務，財團法人保險事業發展中心。
3.林進田，1997，汽車保險損失分配模式化之研究，保險專刊，第50輯，頁65-82。
4.林麗芬 黃文瀚，2003，混合長尾損失分配的尾部純保費估計，保險專刊19卷第一期，頁23-40。
5.凌氤寶，1999，實用產物保險精算，偉仁行。
6.張金裕、李美杏1995，汽車保險費率分類模式之較佳估計，保險專刊，第42輯，頁99-111。
7.梁正德，1993，再談我國汽車保險從人因素係數之釐訂，保險專刊，第32輯，頁176-183。
8.陳強、梁正德及謝良瑾，1992，汽車保險費率從人因素之探討，保險專刊，第30輯，頁97-111。
9.陳強，1992，三十萬強制汽車第三人責任保險保費之計算，保險專刊，第27輯，頁97-111。
10.孫元璋，1998，汽車險人因素費率的貝氏估計與傳統估計之比較，逢甲大學，統計與精算研究所碩士論文。
11.蔡文慈，1996，汽車車體損失險自負額制度及費率釐訂之探討，政治大學保險研究所碩士論文。
12.謝良瑾、吳芬芬、方明川、陳強，1994，汽車險費率從人因素之探討，財政部保險司研究報告8302。
13.賴曜賢，1999，實用財產及責任保險費訂定原理，三民書局。
14.蕭銘宏，1997，貝氏估計法在汽車險從人因素費率係數的應用，逢甲大學，統計與精算研究所碩士論文。
15.Bailey ,R.A. and L.J. Smion,(.1960), “Two Studies in Automobile Insurance Ratemaking,” Proceedings of the Casualty Actuarial Society, Vo1.47(1),1-11.
16.Bailey , R.A. (1963), “Insurance Rates with Minimum Bias, Casualty Actuarial Society ,4-11.
17.Brown , R.L.(1998), “Minimum Bias with Generalized Linear Models,” Proceedings of Casualty Actuarial Society,Vo1.PP187-217.
18.Buhlmann , H.(1967), “Experience Rating and Credibility”, The ASTIN Bulletin, Vol.4( 3), 199-207.
19.David. B. S.(2001), “Measurement of the Effect of Classification Factor Changes in Complex Rating Plans,”Casualty Actuarial Society,222-250.
20.Doerpinghaus, H. I(1993), “The Economics and Politics of Automobile Insurance Rate Classification,” The Journal of Risk and Insurance, 60(1),59-84.
21.Doherty, N.A.(1981), “Is Rate Classification Profitable ?”,The Journal of Risk and Insurance,(June),286-295.
22.Fairley , W.B. , T.J. Tomberlin and H.I. Weisberg(1981), “Pricing Automobile Insurance under a cross-classification of Risks :Evidence from New Jersey,” The Journal of Risk and Insurance, (Sept),505-514.
23.Finger, R.J(1976), “Estimating Pure Premium by Layer-An Approach,” Proceedings of Casualty Actuarial Society, 62. 34-36.
24.Kollar , J.J(1980), “Ratemaking for the personal Automobile Physical Damage Coverage,” Pricing Property and Casualty Insurance Products, discussion paper presented at San Juan, Puerto Rico, Casualty Actuarial Society, 189-190.
25.Mayerson , A.L; D. Jones and N.L Bowers (1968), “On the Credibility of the pure premium,” Proceedings of Casualty Actuarial Society,55,175-185.
26.Mayerson , A.L(1964), “A Bayesian View of Credibility,” Proceedings of Casualty Actuarial society , Vo1.51,85-103.
27.Nasser H.(2003),“Classification Ratemaking Using Decision Trees,” Casualty Actuarial Society,253-284
28.Weiberg, H. And T.J. Tomberlin(1982), “A statistical perspective on Actuarial Methods for Estimating, Pure Premiums from Cross-classified Data,”The Journal of Risk and Insurance, (Dec),30-563.
29.Witt. R. C.(1974), “Credibility standards and pricing in Automobile Insurance,” The Journal of Risks and Insurance,(Sept),375-395.