臺灣博碩士論文加值系統

於保險市場逐步開放之際，如何使經營績效不佳的保險業者能平穩退出市場且不造成市場動盪，即應確實掌控保險業者的清償能力。保險業的清償能力一直是保險監理機關所關切的重點，有效的保險業清償能力預警系統，可於事前即早發現具財務危機的保險公司，監督並輔導改善其財務問題，也能於事後妥善處理，以降低諸多社會成本。為求即早發現問題保險公司，許多學者提出多種預警模型，使用模型諸如：多變量線性區別分析、因素分析、羅吉斯迴歸及類神經網路等。
本文將運用基因演算法建構一保險公司失卻清償能力的預警系統，因國內保險業缺乏破產經驗，故以美國壽險業為樣本進行分析。此外，以基因演算法實證結果與羅吉斯迴歸之結果作一比較，就預測破產公司而言，基因演算法的結果較佳，而對於預測健全公司的部份，兩種模型的結果差異不大，整體而言，基因演算法的預測準確度較羅吉斯迴歸好。

The solvency of insurance company plays an important role in society and has been the focus of insurance regulation. An effective early warning system is conducive to the insurance regulation to detect the insolvent insurance company before they are liquidated. Many Scholars use different models to predict insurer insolvencies. Those models are factor analysis, multivariate linear discriminant analysis, logistic analysis, non-parametric model, recursive partitioning, and neural networks analysis.
We use a genetic algorithm to construct an early warning system for life insurance industry. We use the financial data of USA because of there is no financial data of insolvent insurer of Taiwan. We compared the result of genetic algorithms with the result of logistic regression. With the result of the genetic algorithm and logistic regression model, we find that genetic algorithm performs better in classifying insolvent insurers than logistic regression model, and the result of genetic algorithm in classifying solvent insurers is similar to logistic regression model. Our overall results indicate that the prediction ability of genetic algorithm is better than the logistic regression.

中文摘要i
ABSTRACTii
誌謝iii
Table of Contentiv
List of Tablevi
List of Figureviii
1. Introduction1
2. Genetic Algorithms3
2.1. Basic structure of Genetic Algorithm3
2.2. Genetic Algorithm-Operators6
2.2.1. Selection Operation6
2.2.2. Crossover Operation7
2.2.3. Mutation Operation9
3. Variable Selection and Sample Selection11
3.1. Variable Selection11
3.2. Sample Selection14
4. Genetic Algorithms and Insolvency Risk Analysis15
4.1. Model 1: Genetic Liner Score Function (GLS) with Unclassified Variable15
4.2. Model 2: Genetic Liner Score Function (GLS) with Classified Variables16
4.3. Model 3: Genetic Score by Rules (GSR) with Unclassified Variables18
4.4. Model 4 : Genetic Score by Rules (GSR) with Classified Variables20
5. Result23
5.1. Model 1: Genetic Liner Score Function (GLS) with Unclassified Variables23
5.2. Model 2: Genetic Liner Score Function (GLS) with Classified Variables25
5.3. Model 3: Genetic Score by Rules (GSR) with Unclassified Variables27
5.4. Model 4: Genetic Score by Rules (GSR) with Classified Variables29
5.5. Logistic Regression Models31
5.5.1. None Selection32
5.5.2. Forward Selection34
5.6. Comparison between Genetic Algorithm and Logistic Regression Models35
6. Conclusions39
Reference40
Appendix44

林建智、王儷玲，民國九十年，美國保險業財務分西及清償能力追蹤系統之研究與建議，財團法人保險事業發展中心。
周培之，民國八十四年，遺傳演算法解混合離散型最佳化問題，中山大學機械工程研究所博士論文。
柯俊良，民國八十二年，壽險業清償能力預警模型—類神經網路模型之應用，台灣大學財務金融研究所碩士論文。
陳崇賢，民國八十五年，壽險公司清償能力之研究-遺傳進化之類神經網路的應用，台灣大學財務金融研究所碩士論文。
陳思竹，民國九十一年，演化型再保險自留決策，高雄第一科技大學風險管理與保險研究所碩士論文。
蕭榮興，民國八十八年，股價預測模式中變數選取方法之研究，屏東科技大學資訊管理研究所碩士論文。
魏佑珊，民國九十一年，產險公司破產預測之分析：運用新類神經網路方法，政治大學風險管理與保險研究所碩士論文。
蘇木春、張孝德，民國八十六年，機器學習：類神經網路、模糊系統以及基因演算法則，全華科技出版。
Allen, F., R. Karjalainen, 1999, Using Genetic Algorithms To Find Technical Trading Rules, Journal Of Financial Economics, vol.51, pp. 245–271.
Arifovic, J., 1996, The Behavior of the Exchange Rate in the Genetic Algorithm and Experimental Economies, Journal of Political Economy, Vol.104, no.3, pp. 510-541.
Birchenhall, G., N. Kastrinos, and S. Metcalfe, 1997, Genetic Algorithms in Evolutionary Modeling, Journal of Evolutionary Economics, Vol.7, pp. 375-393.
Beckenbach, F., 1998, Learning by Genetic Algorithms in Economics? , Computational Techniques for Modeling Learning in Economics, pp. 73-100.
Carson, J. M., R. E. Hoyt, 1995, Life Insurer Financial Distress: Classification Models and Empirical Evidence, Journal of Risk and Insurance, Vol.62, pp. 764-775.
Frenken, K., L. Marengo, and M. Valente, 1998, Interdependencies, Nearly-Decomposability and Adaptation, Computational Techniques for Modeling Learning in Economics, pp. 145-165.
Gardner, R. M., S. Elwell, R. Thalman, and E. Rivera, 2000, Solving Tough Semiconductor Manufacturing Problems Using Data Mining, IEEE/SEMI Advanced Semlconductor Conference, pp. 46-55.
Kingdom, J., K. Feldman, 1995, Genetic Algorithms and Application to Finance, Applied Mathematical Finance, Vol.2, pp. 89-116.
Marks, R., H. Schnabl, 1998, Genetic Algorithms and Neural Networks : A Comparison Based on the Prisoners Dilemma, Computational Techniques for Modeling Learning in Economics, pp. 197-219.
Price, T. C., 1997, Using Co-Evolutionary Programming to Simulate Strategic Behavior in Markets, Journal of Evolutionary Economics, Vol.7, pp. 219-254.
Riechmann, T., 1999, Learning and Behavioral Stability An Economic Interpretation of Genetic Algorithms, Journal of Evolutionary Economics, Vol.9, pp. 225-242.
Tam, K. Y., M. Y. Kiang, 1992, Managerial Applications of Neural Networks : The Case of Bank Failure Preditions, Management Science, Vol.38, pp. 926-947.
Varetto, F., 1998, Genetic Algorithms Applications in Analysis of Insolvency Risk, Journal of Banking & Finance, vol.22, pp. 1421-1439.
Xia, Y., B. Liu, S. Wang, and K. K. Lai, 2000, A Model for Portfolio Selection with order of Expected Returns, Computers & Operations Research, vol.27, pp. 409-422.