【作者】 付强；

【导师】 金雪军；

【作者基本信息】 浙江大学 ， 经济学， 2005， 博士

【摘要】 本文针对单个银行失败展开定性和定量的研究,分析了银行失败的原因、识别预测以及处理应对。本文的重点是银行失败的因素识别和失败预测研究。银行失败的因素识别分析可以帮助政府部门和银行界找准问题的所在,采取有效措施防患于未然。预测分析则可以提前发现将要失败的银行,及时做出救助或者关闭的决定,使其负面效应最小化。全文内容安排如下: 第1章说明本文的研究背景、选题意义,评述国内外对银行失败的研究现状,并阐述了本文的创新点和学术价值。 第2章界定银行失败概念并从银行体系内在的不稳定性、市场结构/力量、银行的公司治理、银行效率等角度剖析银行失败的原因。 第3章为银行失败的识别与预测方法研究,研究了各种银行失败预测的模型和方法,包括监管当局的银行失败早期预警体系EWS、外部评级机构的银行评级和各种数量模型方法(线形概率模型、判别分析、Logit分析、Probit分析、COX比例风险模型、非参数特征识别模型以及事件研究法等)。 第4章为使用Logit模型进行单个银行失败估计、识别和预测的完整分析。本章根据1997—2000年间印度尼西亚、马来西亚、韩国和菲律宾四国银行数据,研究银行失败的因素,以决定和预测当恶劣条件发生时什么样的银行更有可能免于失败;对比分析仅仅使用单个银行微观财务数据和使用融合微宏观数据的模型的识别、预测能力的差别。研究发现具有良好的盈利和资本的银行更不可能失败,由于信息不对称的存在,公众无法准确评价银行的状况,一个盈利性强、资本充足的银行往往被公众认为是管理的更好的银行,他们把盈利性和资本看作银行良好性的重要指标,在这两方面都有优势的银行因而也就更有可能在危机中生存下来。使用融合微宏观数据模型的识别、预测能力强于仅仅使用银行微观财务数据模型的识别和预测能力。 第5章研究银行失败的后果和失败处理,在对银行失败后果严密分析的基础上总结和设计合理的处理规则。失败的处理有救助和退出两大类,具体方法的采用必须与具体的问题相适应,但是在处理失败银行时无论采取何种手段和方法,更多还原

【Abstract】 This paper analyses single bank’s failure qualitatively and quantitatively. Analyses the reason, discrimination, prediction, treatment and countermeasures of bank failure. The emphases of this paper are on the factors discrimination and failure prediction. The reseach into factors dicrimination can help government and banks find the question accurately, and take effective measures to prevent trouble before it happens. Prediction analyses can find the failure bank ahead of time, and make the decision on rescue or market exit to minimize its negative effect. The content of the full paper is arranged as follows:Chapter 1 explains the research background and meaning of this paper, comments the current research situation of bank failure both at home and abroad, and sets forth academic value of this reseach.Chapter 2 defines the concept of bank failure, and analyses the reason on the bank failure from the inherent unstability with the banking system, market structure/power, corporate governance, bank efficiency, etc.Chaper 3 studies the discrimination and prediction method of bank failure. Explores various kinds of bank failure models and methods, including early warning system by supervision authorities, the bank outside rating, and econometrics methods, such as, linear probability model, discriminant analysis, Logit analysis, Probit analysis, Cox PHM, Non Parametric Trait Recognition Model, and Event Studies.Chapter 4 employs Logit model to estimate, discriminate and predict the single bank failure completely, according to the data on the banks in Indonesia, Malaysia, Korea S., and Philippine during 1997-2000. The aim of this study is to find bank failure’s factors that determine which bank can survive in a worse economic system, and to analyse the discrimination and prediction power difference between the model using only single bank’s micro financial data and the model merging macroscopical data and micro financial data. The paper discovers that the bank with good更多还原