臺灣博碩士論文加值系統

本篇論文主要的目的在探討比較混合不同資訊的公司破產預測模型與只使用一種資訊的公司破產預測模型中，何者會有較好的預測能力。本研究的資料期間為1998年1月到2008年的12月。本文主要的發現可匯整如下:一、金融機構在預測公司倒閉風險的時候，不應該將主要重心放在同一種類的資訊下。若過度強調某一種類的資訊時，常常就會忽略掉其他重大的變數。在這種情況下，通常會使金融機構做出偏誤的決策。本研究的結果顯示混合的模型確實會比只使用單一資訊的模型更具有預測能力。二、本研究在預測違約機率的方法中，採用了二種不同的破產預測模型:Binary Logistic Regression(羅吉斯迴歸)和Binary Quantile Regression(二元分量迴歸)。結果顯示在大多數的情況下，Binary Quantile Regression會有較佳的預測能力。

This study examines whether the bankruptcy prediction performance of a hybrid model is better than that of various highbred models. The empirical data include firms which were listed on the Taiwan stock market from January 1998 to December 2008. The major findings could be summarized as follows. First, loan institutions should not overemphasize one source of risk factor information, as this will bias decision-making due to the neglect of other important variables. We show that in general the hybrid model has better bankruptcy prediction performance than the highbred models in this study. Second, compared with the setting of constant loadings, the bankruptcy prediction performance of the hybrid model with dynamic loadings achieve better results in most cases.

ABSTRACT..................................................I
CONTENTS.................................................IV
TABLE OF CONTENTS.........................................V
FIGURE OF CONTENTS.......................................VI
1. Introduction...........................................1
2. Literature Review......................................2
3. Empirical Methods......................................5
3.1 Establishment of the model’s variables...............5
3.1.1 Accounting-based information........................5
3.1.2 Market-based information............................5
3.1.3 Governance-based information........................6
3.2 The hybrid model with constant loadings...............7
3.3 The hybrid model with dynamic loadings................8
4. Validation Methodology................................10
4.1 Cumulative Accuracy Profile Curves (CAP curves)......11
4.2 Receiver Operating Characteristic Curves (ROC curves)12
5. Empirical results.....................................12
5.1 The sample firms and their descriptive statistics....12
5.2 Comparison of hybrid and highbred models.............14
5.2.1 Model estimation results...........................14
5.2.2 Bankruptcy prediction tests........................15
5.3 Comparison of BQR and Logit model....................16
5.3.1 Model estimation results...........................16
5.3.2 Bankruptcy prediction tests........................17
6. Conclusions and Directions for Future Research........18
Reference................................................19

TABLE OF CONTENTS
Table 1 Variable Definition..............................21
Table 2 Descriptive Statistics of the Variables..........22
Table 3 Estimates of the Hybrid & Highbred Logit Models..23
Table 4 Accuracy Ratios of Hybrid Logit versus Highbred Logit Models: In-sample Tests............................24
Table 5 Accuracy Ratios of Hybrid Logit versus Highbred Logit Models: Out-of-Sample Tests........................24
Table 6 Accuracy Ratios of Hybrid BQR versus Hybrid Logit Models: In-sample Tests..................................25
Table 7 Type I and Type II Error Rates of the BQR & Logit Models...................................................25
Table 8 Accuracy Ratios of Hybrid BQR versus Hybrid Logit Models: Out-of-Sample Tests..............................26
Table 9 Type I and Type II Error Rates of the BQR & Logit Models...................................................26


FIGURE OF CONTENTS
Figure 1 Illustration of CAP Curves......................27
Figure 2 Illustration of ROC Curves......................27
Figure 3 The CAP curves for the two tested models: Hybrid Logit versus Highbred Logit models.......................28
Figure 4 The ROC curves for the two tested models: Hybrid Logit versus Highbred Logit models.......................28
Figure 5 The loading estimates across various quantile levels of X1: Hybrid BQR versus Hybrid Logit models......29
Figure 6 The loading estimates across various quantile levels of X2: Hybrid BQR versus Hybrid Logit models......29
Figure 7 The loading estimates across various quantile levels of X3: Hybrid BQR versus Hybrid Logit models......30
Figure 8 The loading estimates across various quantile levels of X4: Hybrid BQR versus Hybrid Logit models......30
Figure 9 The loading estimates across various quantile levels of X5: Hybrid BQR versus Hybrid Logit models......31
Figure 10 The loading estimates across various quantile levels of M1: Hybrid BQR versus Hybrid Logit models......31
Figure 11 The loading estimates across various quantile levels of M2: Hybrid BQR versus Hybrid Logit models......32
Figure 12 The loading estimates across various quantile levels of G1: Hybrid BQR versus Hybrid Logit models......32
Figure 13 The loading estimates across various quantile levels of G2: Hybrid BQR versus Hybrid Logit models......33
Figure 14 The CAP curves for the two tested models: Hybrid BQR versus Hybrid Logit models...........................33
Figure 15 The ROC curves for the two tested models: Hybrid BQR versus Hybrid Logit models...........................34

Altman, E.I. (1968), “Financial ratios, discriminant analysis and the prediction of corporate bankruptcy.” Journal of Finance, 13, 589-609

Baldwin, J., and Glezen, G.W. (1986), “Bankruptcy prediction using quarterly financial statement data” Journal of Accounting, Auditing & Finance, 7, 269-285

Beaver, W.H. (1968), “Market Prices, Financial Ratios, and the Prediction of Failure” Journal of Accounting Research, 6, 71-102

Black, F., and Scholes, M. (1973), “The pricing of options and corporate liabilities” Journal of Political Economy, 7, 637-654

Chaganti, R.S., Mahajan, V., and Sharma, S. (1985), “Corporate board size, composition and corporate failures in retailing industry” Journal of Management Studies, 22, 440-417

Daily, C.M., and Dalton, D.R. (1994), “Corporate governance and the bankrupt firm: and empirical assessment,” Strategic Management Journal, 15, 643-654

Hillegeist, S.A., Keating, E.K., Cram, D.P., and Lundstedt, K.G. (2004), “Assessing the probability of bankruptcy” Review of Accounting Studies, 9, 5-34

Kesner, I.F. (1987), “Directors’ stock ownership and organizational performance: An investigation of Fortune 500 companies,” Journal of Management, 13, 499-508

Kordas, G. (2000), “Binary Regression Quantiles,” Ph. D Thesis, University of Illionis at Urbana-Champaign

Kordas, G. (2002), “Credit scoring using binary quantile regression” Working paper

Koenker, R., and Bassett, G.B. (1978), “Regression Quantiles” Econometrica, 46, 33-50

Lee, T.S., and Yeh, Y.H. (2004), “Corporate governance and financial distress: evidence from Taiwan” Journal of Corporate Governance, 12, 378-388

Lemmon, M.L., and Lins, K.V. (2003), “Ownership Structure, Corporate Governance, and Firm Value: Evidence from the East Asian Financial Crisis” Journal of Finance, 58, 1145-1468

Li Leon, M.Y. (2008), “Establishing a hybrid bankruptcy prediction model with dynamic loadings based on accounting-ratio-based and market-based information using binary quantile regression” Working paper

Loffler, G. (2007), “The complementary nature of ratings and market-based measures of default risk” Journal of Fixed Income, 17, 38-47
Merton, R.C. (1974), “On the pricing of corporate debt: The risk structure of interest rates” Journal of Finance, 29, 449-470

Manski, C. (1975), “Maximum score estimation of the stochastic utility model of choice” Journal of Econometrics, 3, 205-228

Manski, C. (1985), “Semiparametric analysis of discrete response: asymptotic properties of the maximum score estimator” Journal of Econometrics, 27, 313-334

Ohlson, J.A. (1980), “Financial ratios and the probabilistic prediction of bankruptcy” Journal of Accounting Research, 18, 109-131

Vassalou, M., and Xing Y. (2004), “Default risk in equity returns” Journal of Financial, 59(2), 831-868