Logistic Discrete Hazard Model在信用風險上之應用

The Application of Logistic Discrete Hazard Model to Credit Risk

陳彥翰(Yen-Han Chen)

指導教授 ： 王泰昌

國立臺灣大學/管理學院/會計學研究所/碩士(2004年)

https://doi.org/10.6342/NTU.2004.00219

摘要

自1998年第四季到2001年第四季，我國有釵h上市上櫃公司發生財務危機，造成公司的員工、客戶、供應商、債權人、股東均蒙受巨大損失。因此，本研究的目標在於建立有效的違約機率預測模型，量化公司信用風險以提供管理當局、投資人、債權人或其他公司關係人決策所需的資訊。
本研究利用logistic 離散涉險模型（logistic discrete hazard model）建立違約機率預測模型，並使用精準度(accuracy ratio) 與Vuong test比較模型的優劣。主要的實證結果如下：
一、本研究採用的各種模型（包括：會計比率模型、市場變數模型、違約間距模型）的相對預測能力在電子業、營建水泥業、其他各業，並不相同。
二、關於本研究的四類解釋變數：會計比率、市場變數、總經變數、股權變數，本研究發現預測能力較低之解釋變數在釵h情況下仍然具有預測能力較高之解釋變數所沒有的資訊，亦即具有增額資訊的效果。
三、本研究發現包含四類解釋變數之合併模型的精準度(accuracy ratio)幾乎都高於個別模型的精準度，只有在營建水泥業之某種違約定義下，合併模型的估計樣本外精準度(out-of-sample accuracy ratio)略低於會計比率模型。

關鍵字

logistic 離散涉險模型 ； 信用風險 ； 精準度

並列摘要

From the fourth quarter of 1998 to the fourth quarter of 2001, many listed and over-the-counter corporations in Taiwan have fallen into financial crises, resulting in huge losses to the employees, customers, vendors, debt holders, and stockholders. The purpose of this thesis is to build an effective model which predicts default probability and quantifies credit risk for information needed by the management, stockholders, debt holder, or other related persons in decision use.
This thesis uses logistic discrete hazard model to predict default probability, and evaluates the performance of models with the concept of Accuracy Ratio and Vuong test. The major research findings are as follows:
I.The relative predictive power of the accounting ratios model, the market variables model, and the default distance model is not the same for the electronic industry, construction and cement industry, and the rest of others.
II.With respect to the four types of explaining variables: accounting ratios, market variables, macroeconomic variables, and stockholding variables, it is found that in many cases the types with inferior predictive power contain incremental information relative to the best one.
III.It is found that the accuracy ratios of the combined models including the four types of explaining variables are higher than those of the separate models including only one type of explaining variable, except under my first definition of default in the construction and cement industry, in which case the out-of-sample accuracy ratio of the combined model is slightly lower than that of the model of accounting ratios.

並列關鍵字

credit risk ； accuracy ratio ； logistic discrete hazard model

參考文獻

Admati, A., P. Pfleiderer and J. Zechner, 1994, Large Shareholder Activisem, Risk Sharing, and Financial Market Equilibrium, Journal of Political Economy, 102, pp. 1097–1130.

Altman, E. I., 1968, Financial Ratios, Discriminant Analysis and the Prediction of Corporate Bankruptcy, Journal of Finance, 23, pp. 578–609.

Altman, E. I., 1989, Measuring corporation bond mortality and performance, Journal of Finance, 44, pp. 909–922.

Altman, E. I., R. Haldeman and P. Narayanan, 1977, ZETA analysis: A new model to identify bankruptcy risk of corporations, Journal of Banking & Finance, 1, pp. 29–54.

Altman, E. I., G.. Marco, and F. Varetto, 1994, Corporate Distress Diagnosis: Comparisons Using Linear Discriminant Analysis and Neural Networks, Journal of Banking and Finance, 18, pp. 505–529.