1. NCCU Academic Hub
  2. Publications
  3. College of Commerce
  4. Theses
  5. Item

Publications-Theses

Article View/Open

Publication Export

Google ScholarTM

NCCU Library

Citation Infomation

Related Publications in TAIR

題名 導入資料採礦技術於中小企業營造業信用風險模型之建置
Establishment of credit risks model for the construction industry of the SMEs with data mining techniques
作者 謝欣芸
Hsieh, Shin-Yun
貢獻者 鄭宇庭<br>薛惠敏
Cheng, Yu-Ting<br>Hsueh, Huei-Min
謝欣芸
Hsieh, Shin-Yun
關鍵詞 新巴塞爾資本協定
信用風險模型
羅吉斯迴歸
資料採礦
The New Basel Capital Accord
Credit-risks model
Logistic Regression
Data Mining
日期 2009
上傳時間 9-May-2016 15:11:40 (UTC+8)
摘要 為了符合國際清算銀行在 2006 年通過的新巴賽爾資本協定，且有鑑於近年
     來整體經濟環境欠佳，銀行業者面對外部的規定以及內部的需求，積極地尋求
     信用風險模型的建置方法，希望將整個融資的評等過程系統化以提高對信用風
     險的控管。
     本研究希望利用 92 至94 年未上市上櫃中小企業之營造業的資料，依循新
     巴賽爾資本協定之規定並配合資料採礦的技術，擬出一套信用風險模型建置與
     評估的標準流程，其中包含企業違約機率模型以及信用評等系統的建置，前者
     能預測出授信戶的違約情形以及違約機率；後者則是能利用前者的分析結果將
     授信戶分成數個不同的等級，藉此區別授信戶是否屬於具有高度風險的違約授
     信戶，期待能提供銀行業者作為因應新巴賽爾協定中內部評等法的建置，以及
     中小企業的融資業務上內部風險管理的需求一個參考的依據。
     研究結果共選出 5 個變數作為企業違約機率模型建立之依據，訓練資料以
     及原始資料的AUC 分別為0.799 以及0.773，表示模型能有效的預測違約機率
     並判別出違約授信戶以及非違約授信戶。接著，經過回顧測試與係數拔靴測試，
     證實本研究的模型具有一定的穩定性。另外，透過信用評等系統將所有授信戶
     分為8 個評等等級，並藉由等級同質性檢定以及敏感度分析的測試，可以驗證
     出本研究之評等系統具有將不同違約程度的授信戶正確歸類之能力。最後，經
     由轉移矩陣可以發現，整體而言，營造業在2003 年到2005 年間的表現有逐漸
     好轉的趨勢，與營造業實際發展情形相互比較之下，也確實得到相互吻合的結
     論。
In order to conform to the New Basel Capital Accord passing in 2006 by the
     Bank for International Settlements and due to the slump faced by economies
     globally and the rise in the number of defaulters in the recent years, the banking
     industry has aggressively looked for ways to establish the reliable credit risk model
     that can accommodate required regulations set forth by the Accord as well as the
     internal banking procedure demands. The banking industry attempts to standardize
     the process of evaluating credit rating in regards to capital risk in the loan business
     to enhance the control of credit risks.
     The attempt of this research is to perform the process of the establishment and
     evaluation of the credit risk model which includes the default risk model of
     companies and the credit rating system within the framework of the New Basel
     Capital Accord using the statistical tool known as data mining. The data adopted in
     this study is taken from the construction industry of the SMEs from 2003 to 2005.
     The default risk model assesses the probability whether a company is at risk of being
     defaulted. In addition the credit rating system assigns credit scores to a company in
     question based on the application result from the default risk model to differentiate
     those who have high risk of being defaulted. More importantly this research
     provides banking industry of varying degrees of complexity to monitor its risk
     assessment as well as becoming a reference basis of the loan business in the SMEs.
     Based on the result of this study, five variables are selected as the default
     probability model basis. The AUC for the training data is 0.799 and for the raw data
     is 0.773 which represents the accuracy and reliability of the model in predicting the
     probability of default risk and determining the likelihood of the companies to default.
     After series of testing, our model stability plays a key role in determining whether
     the algorithm produces an optimal model in this study. The credit rating system
     formulates credit scores of the companies into 8 credit ratings. Applying
     homogeneity test and sensitive analysis, this study is able to verify the validity and
     accuracy of the rating system to correctly classify different levels of credit risk that
     could have jeopardized the companies to default. Finally, through the transformation
     matrix, there has been an improvement trend of performance in the construction
     industry from 2003 to 2005 which coincides with the result of this study.
參考文獻 一、 中文文獻
     沈大白、賴柏志(2004)，「壓力測試於信用風險模型之應用」，財團法人金融合
     徵信中心，民國九十三年二月號。
     沈中華、林昆立(2007)，「台灣金融機構適足資本之壓力測試」，金融風險管理
     季刊，民國九十六年三月號。
     阮正治、江景清(2004)，「台灣企業信用評分模型建置與驗證」，金融風險管理
     季刊，民國九十三年六月號。
     李永彬(2004)，「新巴賽爾協定與產業研究」，華南金控月刊，第12 期，頁40-42。
     林公韻(2005)，「信用違約機率之預測－Robust Logistic Regression」，國立政治
     大學金融研究所碩士論文。
     林孟寬(2008)，「企業信用評等模型-以營造業為例」，國立政治大學統計研究所
     碩士論文。
     金管會，新巴塞爾資本協定中文版(2004)。
     張炳輝（2002），「新巴賽爾資本協定與信用風險標準法之介紹」，華南金控月刊，
     第五期，頁5-14。
     經濟部中小企業處，中小企業白皮書(2008)。
     華銀信用風險 IRB 組(2004)，「Basel II-信用風險內部評等法簡介」，華銀專題論
     述，頁12-14。
     銀行局，新巴賽爾資本協定簡介(2006)。
     張大成、林郁翎、黃士賓(2008)，「應用拒絕推論技術建構企業財務危機預警模
     型：以台灣上市公司為例」，2008 管理創新與新願景研討會，台灣台北
     張永吉(2005)，「新版巴塞爾資本協定本國銀行業信用風險之探討- CreditRisk+
     信用風險模型研究」，東吳大學商學院企業管理學系，碩士論文。
     詹益宗(2005)，「財務危機預警模型之比較」，交通大學財務金融學系碩士論文
     劉威漢(2004)，「財金風險管理理論、應用與發展趨勢」，智勝文化事業。
     謝邦昌、鄭宇庭、蘇志雄(2009)，「Data Mining 概述-以Clementine12.0 為例」，
     中華資料採礦協會。
     二、 英文文獻
     Altman, E.I.(1968). “Financial Ratios, Discriminant Analysis and the Prediction of
     Corporate Bankruptcy. ” Journal of Finance 23:589-609.
     Beaver, W.H.(1966). “Financial Rations as Predictors of Failure, ” Journal of
     Accounting Research (4): pages 71-111.
     Berry M.J.A. and G..S. Linoff (1997), “Data Mining Techniques: for Marketing,
     Sales, and Customer Support, ” John Wiley & Sons.
     Dutta, S., and, S. Shekhar (1988), “Bond Rating: A Non-Conservative Application of
     Neural Network,” Proceeding of the IEEE International Conference on Neural
     Networks, Ⅱ, pages 443-450.
     Fayyad U.M. (1996), “Data Mining and Knowledge Discovery: Making Sense Out
     of Data,” IEEE Expert, 11(10), pages 20-25.
     Fayyad U.M., P.S. Gregory, P. Smyth and R. Uthurusamy(1996), “Advances in
     Knowledge Discovery and Data Mining, ” The MIT Press.
     Gregory Piatetsky-Shapiro & Frawley W. J.(1991), “Knowledge Discovery in
     Databases, ”The MIT Press, pages 126–140.
     Hosmer, D. W., and S. Lemeshow(2000), “Applied Logistic Regression, ” 2th ed.
     John Wiley & Sons, Inc.
     Jarrow, R., D. Lando, and S. Turnbull(1997), “A Markov model for the term
     structure of credit spread,” Review of Financial Studies 10, pages 481- 523.
     Kleissner, C.(1998), “Data Mining for the Enterprise，Proc. of 31st Hawaii Int’l
     Conf. on System Sciences, ” 7 295-304.
     Martin, D.(1977) “Early Warning of Bank Failure: A Logistic Regression
     Approach,” Journal of Banking and Finance, Vol. 1, pages 249-276.
     Mays, E.(2001), “Handbook of Credit Scoring, Chicago”: Glenlake.
     Merton, R.(1974), “On the pricing of corporate debt：The risk structure of interest
     rates,” Journal of Finance 29, pages 449-470.
     Odom, M.D., and R. Sharda (1990), “A Neural Network Model for Bankruptcy
     Prediction,” Proceedings of the International Joint Conference on Neural
     Networks.
     Ohlson, J.A.(1980). “Financial Ratios and the Probabilistic Prediction of
     Bankruptcy, ” Journal of Accounting Research (Spring): pages 109-131.
     三、 參考網站
     中華信評公司 http://www.taiwanratings.com/tw/
     內政部營建署http://www.cpami.gov.tw/web/index.php
     曾國烈(2005)，信用風險模型簡介
     www.banking.gov.tw/public/Attachment/872115285771.ppt
     行政院金融管理委員會 http://www.fscey.gov.tw/
     行政院主計處 http://www.dgbas.gov.tw/mp.asp?mp=1
     銀行公會 https://www.ba.org.tw/
     經濟部中小企業處 http://www.moeasmea.gov.tw/mp.asp?mp=1
描述 碩士
國立政治大學
統計學系
96354025
資料來源 http://thesis.lib.nccu.edu.tw/record/#G0096354025
資料類型 thesis
dc.contributor.advisor 鄭宇庭<br>薛惠敏zh_TW
dc.contributor.advisor Cheng, Yu-Ting<br>Hsueh, Huei-Minen_US
dc.contributor.author (Authors) 謝欣芸zh_TW
dc.contributor.author (Authors) Hsieh, Shin-Yunen_US
dc.creator (作者) 謝欣芸zh_TW
dc.creator (作者) Hsieh, Shin-Yunen_US
dc.date (日期) 2009en_US
dc.date.accessioned 9-May-2016 15:11:40 (UTC+8)-
dc.date.available 9-May-2016 15:11:40 (UTC+8)-
dc.date.issued (上傳時間) 9-May-2016 15:11:40 (UTC+8)-
dc.identifier (Other Identifiers) G0096354025en_US
dc.identifier.uri (URI) http://nccur.lib.nccu.edu.tw/handle/140.119/95124-
dc.description (描述) 碩士zh_TW
dc.description (描述) 國立政治大學zh_TW
dc.description (描述) 統計學系zh_TW
dc.description (描述) 96354025zh_TW
dc.description.abstract (摘要) 為了符合國際清算銀行在 2006 年通過的新巴賽爾資本協定，且有鑑於近年
     來整體經濟環境欠佳，銀行業者面對外部的規定以及內部的需求，積極地尋求
     信用風險模型的建置方法，希望將整個融資的評等過程系統化以提高對信用風
     險的控管。
     本研究希望利用 92 至94 年未上市上櫃中小企業之營造業的資料，依循新
     巴賽爾資本協定之規定並配合資料採礦的技術，擬出一套信用風險模型建置與
     評估的標準流程，其中包含企業違約機率模型以及信用評等系統的建置，前者
     能預測出授信戶的違約情形以及違約機率；後者則是能利用前者的分析結果將
     授信戶分成數個不同的等級，藉此區別授信戶是否屬於具有高度風險的違約授
     信戶，期待能提供銀行業者作為因應新巴賽爾協定中內部評等法的建置，以及
     中小企業的融資業務上內部風險管理的需求一個參考的依據。
     研究結果共選出 5 個變數作為企業違約機率模型建立之依據，訓練資料以
     及原始資料的AUC 分別為0.799 以及0.773，表示模型能有效的預測違約機率
     並判別出違約授信戶以及非違約授信戶。接著，經過回顧測試與係數拔靴測試，
     證實本研究的模型具有一定的穩定性。另外，透過信用評等系統將所有授信戶
     分為8 個評等等級，並藉由等級同質性檢定以及敏感度分析的測試，可以驗證
     出本研究之評等系統具有將不同違約程度的授信戶正確歸類之能力。最後，經
     由轉移矩陣可以發現，整體而言，營造業在2003 年到2005 年間的表現有逐漸
     好轉的趨勢，與營造業實際發展情形相互比較之下，也確實得到相互吻合的結
     論。		zh_TW
dc.description.abstract (摘要) In order to conform to the New Basel Capital Accord passing in 2006 by the
     Bank for International Settlements and due to the slump faced by economies
     globally and the rise in the number of defaulters in the recent years, the banking
     industry has aggressively looked for ways to establish the reliable credit risk model
     that can accommodate required regulations set forth by the Accord as well as the
     internal banking procedure demands. The banking industry attempts to standardize
     the process of evaluating credit rating in regards to capital risk in the loan business
     to enhance the control of credit risks.
     The attempt of this research is to perform the process of the establishment and
     evaluation of the credit risk model which includes the default risk model of
     companies and the credit rating system within the framework of the New Basel
     Capital Accord using the statistical tool known as data mining. The data adopted in
     this study is taken from the construction industry of the SMEs from 2003 to 2005.
     The default risk model assesses the probability whether a company is at risk of being
     defaulted. In addition the credit rating system assigns credit scores to a company in
     question based on the application result from the default risk model to differentiate
     those who have high risk of being defaulted. More importantly this research
     provides banking industry of varying degrees of complexity to monitor its risk
     assessment as well as becoming a reference basis of the loan business in the SMEs.
     Based on the result of this study, five variables are selected as the default
     probability model basis. The AUC for the training data is 0.799 and for the raw data
     is 0.773 which represents the accuracy and reliability of the model in predicting the
     probability of default risk and determining the likelihood of the companies to default.
     After series of testing, our model stability plays a key role in determining whether
     the algorithm produces an optimal model in this study. The credit rating system
     formulates credit scores of the companies into 8 credit ratings. Applying
     homogeneity test and sensitive analysis, this study is able to verify the validity and
     accuracy of the rating system to correctly classify different levels of credit risk that
     could have jeopardized the companies to default. Finally, through the transformation
     matrix, there has been an improvement trend of performance in the construction
     industry from 2003 to 2005 which coincides with the result of this study.		en_US
dc.description.tableofcontents 第壹章 緒論 ............................................................................................................ 1
     第一節研究背景與動機................................................................................................. 1
     第二節研究目的............................................................................................................. 3
     第三節研究架構............................................................................................................. 3
     第貳章 文獻探討 .................................................................................................... 5
     第一節新巴塞爾資本協定簡介..................................................................................... 5
     第二節信用風險模型及信用評等介紹......................................................................... 9
     第三節資料採礦標準流程........................................................................................... 14
     第四節相關文獻探討................................................................................................... 17
     第參章 研究方法 .................................................................................................. 22
     第一節研究對象........................................................................................................... 22
     第二節研究流程........................................................................................................... 22
     第三節研究方法........................................................................................................... 26
     第肆章 實證分析 .................................................................................................. 43
     第一節資料處理與模型變數選取............................................................................... 43
     第二節建置預測模型................................................................................................... 47
     第三節機率校準與評等等級....................................................................................... 54
     第四節模型的評估與驗證........................................................................................... 57
     第伍章 結論與建議 .............................................................................................. 67
     第一節研究結果........................................................................................................... 67
     第二節研究建議........................................................................................................... 69
     參考文獻 ...................................................................................................................... 72
     附錄 .............................................................................................................................. 75		zh_TW
dc.source.uri (資料來源) http://thesis.lib.nccu.edu.tw/record/#G0096354025en_US
dc.subject (關鍵詞) 新巴塞爾資本協定zh_TW
dc.subject (關鍵詞) 信用風險模型zh_TW
dc.subject (關鍵詞) 羅吉斯迴歸zh_TW
dc.subject (關鍵詞) 資料採礦zh_TW
dc.subject (關鍵詞) The New Basel Capital Accorden_US
dc.subject (關鍵詞) Credit-risks modelen_US
dc.subject (關鍵詞) Logistic Regressionen_US
dc.subject (關鍵詞) Data Miningen_US
dc.title (題名) 導入資料採礦技術於中小企業營造業信用風險模型之建置zh_TW
dc.title (題名) Establishment of credit risks model for the construction industry of the SMEs with data mining techniquesen_US
dc.type (資料類型) thesisen_US
dc.relation.reference (參考文獻) 一、 中文文獻
     沈大白、賴柏志(2004)，「壓力測試於信用風險模型之應用」，財團法人金融合
     徵信中心，民國九十三年二月號。
     沈中華、林昆立(2007)，「台灣金融機構適足資本之壓力測試」，金融風險管理
     季刊，民國九十六年三月號。
     阮正治、江景清(2004)，「台灣企業信用評分模型建置與驗證」，金融風險管理
     季刊，民國九十三年六月號。
     李永彬(2004)，「新巴賽爾協定與產業研究」，華南金控月刊，第12 期，頁40-42。
     林公韻(2005)，「信用違約機率之預測－Robust Logistic Regression」，國立政治
     大學金融研究所碩士論文。
     林孟寬(2008)，「企業信用評等模型-以營造業為例」，國立政治大學統計研究所
     碩士論文。
     金管會，新巴塞爾資本協定中文版(2004)。
     張炳輝（2002），「新巴賽爾資本協定與信用風險標準法之介紹」，華南金控月刊，
     第五期，頁5-14。
     經濟部中小企業處，中小企業白皮書(2008)。
     華銀信用風險 IRB 組(2004)，「Basel II-信用風險內部評等法簡介」，華銀專題論
     述，頁12-14。
     銀行局，新巴賽爾資本協定簡介(2006)。
     張大成、林郁翎、黃士賓(2008)，「應用拒絕推論技術建構企業財務危機預警模
     型：以台灣上市公司為例」，2008 管理創新與新願景研討會，台灣台北
     張永吉(2005)，「新版巴塞爾資本協定本國銀行業信用風險之探討- CreditRisk+
     信用風險模型研究」，東吳大學商學院企業管理學系，碩士論文。
     詹益宗(2005)，「財務危機預警模型之比較」，交通大學財務金融學系碩士論文
     劉威漢(2004)，「財金風險管理理論、應用與發展趨勢」，智勝文化事業。
     謝邦昌、鄭宇庭、蘇志雄(2009)，「Data Mining 概述-以Clementine12.0 為例」，
     中華資料採礦協會。
     二、 英文文獻
     Altman, E.I.(1968). “Financial Ratios, Discriminant Analysis and the Prediction of
     Corporate Bankruptcy. ” Journal of Finance 23:589-609.
     Beaver, W.H.(1966). “Financial Rations as Predictors of Failure, ” Journal of
     Accounting Research (4): pages 71-111.
     Berry M.J.A. and G..S. Linoff (1997), “Data Mining Techniques: for Marketing,
     Sales, and Customer Support, ” John Wiley & Sons.
     Dutta, S., and, S. Shekhar (1988), “Bond Rating: A Non-Conservative Application of
     Neural Network,” Proceeding of the IEEE International Conference on Neural
     Networks, Ⅱ, pages 443-450.
     Fayyad U.M. (1996), “Data Mining and Knowledge Discovery: Making Sense Out
     of Data,” IEEE Expert, 11(10), pages 20-25.
     Fayyad U.M., P.S. Gregory, P. Smyth and R. Uthurusamy(1996), “Advances in
     Knowledge Discovery and Data Mining, ” The MIT Press.
     Gregory Piatetsky-Shapiro & Frawley W. J.(1991), “Knowledge Discovery in
     Databases, ”The MIT Press, pages 126–140.
     Hosmer, D. W., and S. Lemeshow(2000), “Applied Logistic Regression, ” 2th ed.
     John Wiley & Sons, Inc.
     Jarrow, R., D. Lando, and S. Turnbull(1997), “A Markov model for the term
     structure of credit spread,” Review of Financial Studies 10, pages 481- 523.
     Kleissner, C.(1998), “Data Mining for the Enterprise，Proc. of 31st Hawaii Int’l
     Conf. on System Sciences, ” 7 295-304.
     Martin, D.(1977) “Early Warning of Bank Failure: A Logistic Regression
     Approach,” Journal of Banking and Finance, Vol. 1, pages 249-276.
     Mays, E.(2001), “Handbook of Credit Scoring, Chicago”: Glenlake.
     Merton, R.(1974), “On the pricing of corporate debt：The risk structure of interest
     rates,” Journal of Finance 29, pages 449-470.
     Odom, M.D., and R. Sharda (1990), “A Neural Network Model for Bankruptcy
     Prediction,” Proceedings of the International Joint Conference on Neural
     Networks.
     Ohlson, J.A.(1980). “Financial Ratios and the Probabilistic Prediction of
     Bankruptcy, ” Journal of Accounting Research (Spring): pages 109-131.
     三、 參考網站
     中華信評公司 http://www.taiwanratings.com/tw/
     內政部營建署http://www.cpami.gov.tw/web/index.php
     曾國烈(2005)，信用風險模型簡介
     www.banking.gov.tw/public/Attachment/872115285771.ppt
     行政院金融管理委員會 http://www.fscey.gov.tw/
     行政院主計處 http://www.dgbas.gov.tw/mp.asp?mp=1
     銀行公會 https://www.ba.org.tw/
     經濟部中小企業處 http://www.moeasmea.gov.tw/mp.asp?mp=1		zh_TW