Balance-Sheet-Based or Equity-Based Credit Scoring: A Dynamic Perspective Using Binary Quantile Regression

• Main Authors: Hsin-Hsin Huang, 黃心欣
• Other Authors: Ming-Yuan Li
• Format: Others
• Language: en_US
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/01220511317462716264

碩士 === 國立成功大學 === 財務金融研究所 === 96 === Financial data for U.S. firms (bankrupt/healthy firms) listed during 1996-2006 are analyzed. A key feature of this study is analysis of changing distribution of corporate nature across firms and over time by binary quantile regression (hereafter BQR) model and comparison of the results with Logistic Regression (hereafter LR). The nonlinearities derived from conditional BQR reveal considerable discrimination between different sources of credit scoring indicator vary by firms-nature (bankrupt/ healthy firms). On view of pre-quarter prediction, source from market indicator have interpretation on bankrupt firms, source form accounting indicator has less interpretation on bankrupt firms. By validation of credit model, we evaluating BQR and LR performance have separate goodness. BQR model at each quantile outperform on type I error. For type II error, LR model outperform than BQR. In view of investor risk preference, BQR model is suitable for risk-averse investor.