鍾經樊Chung, Ching-Fan臺灣大學:經濟學研究所陳韋達Chen, Wei-DaWei-DaChen2010-05-052018-06-282010-05-052018-06-282009U0001-2407200916212500http://ntur.lib.ntu.edu.tw//handle/246246/179548本文欲建立銀行企業金融放款組合資產之損失分配, 以多因子模型建立放款對象違約之間的相關性, 其中股價報酬率為觸發變數, 當股價報酬率低於某個門檻的時候,放款對象會發生違約。藉由觸發變數受到總體因子的系統性影響, 同時設計受總體因子影響的結構式LGD 模型, 因子模型引入違約機率(probability of default,PD) 及違約損失率(loss given default, LGD) 的相關。為了得到動態經濟資本規劃,本文進一步修改總體因子設定, 以向量自我迴歸模型(vector autoregression, VAR) 進行總體因子預測, 經由蒙地卡羅模擬, 以我國上市櫃的 537 間公司放款作為虛擬組合資產, 求得給定總體條件下的信用組合資產損失分配。以給定總體因子條件下的損失分配做經濟資本的規劃, 若景氣好需要較少的資本, 銀行可以承做多一些較風險的曝險以增加獲利;反之景氣差時則增加計提資本,作為景氣差時的防備。最後, 對信用損失分配進行總體因子受到衝擊時的敏感度分析, 以得知損失分配之預期損失、風險值、經濟資本在衝擊之下將會如何改變。結果顯示在總體因子受到衝擊下, 銀行放款的損失是非對稱的。在受到不利衝擊時, 預期損失通常會增加, 經濟資本的改變則需視預期損失和風險值相對增加的比例而定。We construct a loss distribution of corporate loan portfolio and use multifactor model to construct default correlation between obligors. The equity return of eachbligor is the trigger variable. When the obligor’s equity return is below a threshold, the obligor will default. Under the systematic effect of macroeconomic factorsnd the structural LGD model, the probability of default (PD) of each obligor is correlated with his loss given default (LGD). n order to obtain dynamic economic capital adjusted with macroeconomic circumstances,e apply vector autoregressive model in predicting macroeconomic factors and use Monte Carlo simulation to generate a conditional portfolio loss of 537 Taiwanese firms loan portfolio conditional on macroeconomic circumstances. By using conditional loss distribution in planning economic capital, banks can lend more and earn more money during the period of economic expansion and lend less to keep more capital during the period of economic distress. Also, we conduct scenario analysis and find how the expected loss, unexpected loss, VaR, and economic capital respond to these shocked scenarios. Under these shocked scenarios, the change of expected loss is unsymmetrical. In the case of unfavorable scenario, the expected loss will increase, and the change of economic capital will depend on the relative changes of VaR and expected loss.目錄 前言1 文獻回顧3.1 損失變數. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 違約曝險額. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 違約機率模型. . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4 違約損失率模型. . . . . . . . . . . . . . . . . . . . . . . . . . . 11.5 損失分配. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.6 經濟資本. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 違約損失模型16.1 違約機率. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.2 違約損失率. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.3 違約曝險額. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26.4 敏感度分析與資本適足率. . . . . . . . . . . . . . . . . . . . . . 27 組合資產信用損失模型實證結果29.1 蒙地卡羅模擬基本設定. . . . . . . . . . . . . . . . . . . . . . . 29.2 信用損失模擬結果. . . . . . . . . . . . . . . . . . . . . . . . . . 35 結論 42考文獻 45錄1 48錄2 49錄3 50application/pdf1010982 bytesapplication/pdfen-US風險值信用組合資產管理向量自我迴歸敏感度分析經濟資本違約相關Value at Risk (VaR)Credit PortfolioManagementVector AutoRegressionScenario AnalysisEconomic CapitalDefault Correlationthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/179548/1/ntu-98-R96323035-1.pdf