Comparisons of Bankrupt Models between Fixed and Time-varying Risk Factors

• Main Authors: Yi-Ling Chen, 陳怡伶
• Other Authors: Shu-Hui Yu
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/10658795205475827928

碩士 === 國立暨南國際大學 === 財務金融學系 === 92 === Bankrupt models have been receiving considerable attention from researchers and practitioners for the past several decades. In the literature, multivariate discriminant analysis, logistic regression models, and conditional probability models are frequently used for analyzing firm’s bankruptcy. Unfortunately, these models are static in nature and difficult to model the dynamics of bankruptcy. Shumway (2001) proposed a hazard model to fill this gap. However, since the bankruptcy risk can be time-varying, in this paper, we take this factor into account and proposed a new dynamic hazard model. Comparisons between the proposed model and Shumway''s fixed risk factor hazard model are also performed. First, we analyze firms in the UK industrial sector from two different viewpoints and relate risk factors to financial theory. Further, we clarify how the correlations between data influence model’s predictive power. Our findings confirm that, optimal capital structure hypothesis and size are the most explanatory variables in fixed risk factors structure. In contrast to time-varying risk factors structure, insolvency hypothesis is more important in predicting failure rate. In addition, we also show that the predictive power of the time-varying risk model is twice higher than the predictive power of Shumway’s model. Finally, we point out that Shumway’s basic assumption on hazard model, that is, hazard model is not sensitive to dependence between data, could be erroneous.