新版巴塞爾協定已於2004年6月底正式定案，為了因應世界潮流，台灣也將開始實行新版巴塞爾協定，而在當今信用風險評價模型中的結構式模型裡，以Moody's KMV 模型最被廣泛使用。但是台灣的經濟體系較特殊，主要是以中小企業為主，欠缺市場資料，故僅能用 Moody's KMV 公司發展的 private firm model (PFM) 計算違約機率以量化信用風險。然而 PFM 有其缺失且應用於國內中小企業的相關文獻大多效果不彰，故本研究的目的在於改進 Moody's KMV 模型應用於未上市、未上櫃公司之方法，除了在股價的隨機過程中加入跳躍擴散模型外，更利用類神經網路模型估計未上市、未上櫃公司的股價資訊，並利用蒙地卡羅模擬法模擬公司資產價值路徑，以期更精確地量化信用風險。為了檢測類神經網路模型的能力，本研究分別對上市公司與未上市、未上櫃公司做實證，並用累積正確率曲線比較不同模型鑑別力的好壞。研究結果顯示，無論是在上市公司部分或未上市、未上櫃公司部分，本研究模型的鑑別力均比 Moody's KMV模型佳，並且上市公司部分與未上市、未上櫃公司部分得到的實證結果一致，顯示類神經網路模型估計股價的能力良好，相信本研究模型將能提供國內銀行於計算中小企業信用風險時，另一種值得參考之方法。

The new version of Basel Capital Accord has been formally rectified in the end of June, 2004. In order to follow the worldwide trend, the new version will also be executed in Taiwan. Among all contemporary structural modules of credit risk assessing models, the most prevalently implemented model is the KMV model. However, because the economic system in Taiwan, which is mainly based on small and medium enterprises and thus is deprived of market information, is unique, private firm model (PFM) developed by Moody's KMV Company is therefore employed to calculate the probability of default (PD) so as to quantify credit risk. However, private firm model has a lot of defects and relevant literature has shown that the application of PFM is not effective. Thus, the purpose of the current paper is to improve the method of applying Moody's KMV on evaluating non-listed companies. This paper not only adds jump-diffusion model to stock random process but also uses neural network to evaluate the stock price of non-listed companies. Furthermore, the present paper also adopts Monte Carlo simulation to simulate asset value paths in an attempt to more accurately quantify credit risks. The results show that our model has better discriminatory power in terms of both listed and non-listed companies; moreover, the empirical finding on both types of companies is in higher jump-coefficients. It can be concluded that the capacity of neural network in evaluating stock prices is prominent. Thus, it is believed that the current research will provide the domestic banks with another recommendable method to assess the credit risk of small and medium enterprises.

[1] 葉怡成 (2002)， 《應用類神經網路》 ，台北：儒林圖書有限公司。
[2] Beneish, M. and E. Press (1995), “Interrelation among events of default,” Working paper.
[3] Black, F. and M. Scholes (1973), “The Pricing of Options and Corporate Liabilities,” Journal of Political Economy, 81, 637-659.
[4] Black, F. and J. C. Cox (1976), “Valuing Corporate Securities: Some Effects of Bond Indenture Provisions,” Journal of Finance, 31, 351-367.
[5] Brennan, M. and E. Schwartz (1978), “Corporate Income Taxes, Valuation, and the Problem of Optimal Capital Structure,” Journal of Business, 51,103-114.
[6] Chiang, W. C. et al. (1996), “A Neural Network Approach to Mutual Fund Net Asset Value Forecasting,” Omega, 24, 205-215.
[7] Duffie, D. and D. Lando (1997), “Term Structure of Credit Spreads with Incomplete Accounting Information,” Working paper.
[8] Dutta, S. and S. Shekhar (1988), “Bond Rating: A Non-Conservative Application of Neural Networks,” IEEE International Conference on Neural Networks, San Diego, l2, 443-450.
[9] Desai, V. S. and R. Bharati (1998), “The Efficacy of Neural Networks in Predicting Returns on Stock and Bonds Indices,” Decision Sciences Journal, 29, 405-426.
[10] Engelmann et al. (2003), “Testing Rating Accuracy,” Risk, 82-86.
[11] Fons, J. S. (1994), “Using Default Rates to Model the Term Structure of Credit Risk,” Financial Analysts Journal, 25-32.
[12] Franks, J. and W. Torous (1989), “An empirical investigation of U.S. firms in reorganizations,” Journal of Finance, 44, 747-769.
[13] Jorion, P. (1988), “On Jump Processes in the Foreign Exchange and Stock Markets,” The Review of Financial Studies, 1, 427-445.
[14] Kanas, A. and A. Yannopoulos (2001), “Comparing Linear and Non-Linear Forecasts for Stock Return,” International Review of Economics and Finance, 10, 383-398.
[15] Kim, I. J. et al. (1989), “The Valuation of Corporate Fixed Income Securities,” Working paper.
[16] Leland, H. (1994), “Bond Prices, Yield Spreads, and Optimal Capital Structure with Default Risk,” Working paper.
[17] Liu, W. et al. (2000), “Sovereign Debt Service Capacity Estimation by Logistic Regression and Neural Networks,” Multinational Business Review, 8, 51-57.
[18] Longstaff, F. A. and E. S. Schwartz (1995), “A Simple Approach to Valuing Risky Fixed and Floating Rate Debt,” Journal of Finance, 3, 789-819.
[19] Mason, S. and S. Bhattacharya (1981), “Risky Debt, Jump Processes and Safety Covenants,” Journal of Financial Economics, 9, 281-307.
[20] Merton, R. (1974), “On the Pricing of Corporate Debt: The Risk Structure of Interest Rates,” Journal of Finance, 28, 449-470.
[21] Merton, R. (1976), “Option Pricing when Underlying Stock Returns are Discontinuous,” Journal of Financial Economics, 3,125-144.
[22] Moshiri, S. and N. Cameron (2000), “Neural Network versus Econometric Models in Forecasting Inflation,” Journal of Forecasting, 19, 201-217.
[23] Nguyen, N. and A. Cripps (2001), “Predicting Housing Value: A Comparison of Multiple Regression Analysis and Artificial Neural Networks,” Journal of Real Estate Research, 22, 313-336.
[24] Rumelhart, D. E. et al. (1986), “Learning Internal Representation by Error Propagation,” in D. Rumelhart and J. McClelland, (ed.), Parallel Distributed Processing, 318-362, MIT Press.
[25] Sarig, O. and A. Warga (1989), “Some Empirical Estimates of the Risk Structure of Interest Rates,” Journal of Finance, 44, 1351-1360.
[26] Shaaf, M. (2000), “Predicting Recession Using the Yield Curve: An Artificial Intelligence and Econometric Comparison,” Eastern Economic Journal, 26, 171-190.
[27] Zhou, C. (1997), “A Jump-Diffusion Approach to Modeling Credit Risk and Valuing Defaultable Securities,” Working paper.
[28] Zhou, C. (2001), “The Term Structure of Credit Spreads with Jump Risk,” Journal of Banking and Finance, 25, 2015-2040.