基于SVM的中国商业银行危机预警模型研究

发布时间：2018-01-26 13:59

本文关键词： 商业银行危机预警支持向量机遗传算法　出处：《大连理工大学》2013年硕士论文　论文类型：学位论文

【摘要】：危机预警是商业银行经营领域的经典且持续演进的重要命题。在2008年的金融危机背景下,如何识别银行危机特征,揭示银行危机发生机理,从而构建准确有效的预警模型等问题已日益引起理论界与实践界的高度重视。本文首先从银行危机的界定、危机产生原因以及危机预警的指标和方法这四个方面入手,系统地回顾了国内外对银行危机预警研究的相关文献,进而构建了本文的理论框架——中国商业银行危机的诱因分析以及基于支持向量机的预警模型。并在此基础上考虑了商业银行危机产生的内在根源、外来威胁,分析了中国商业银行危机的诱发机理并确立了预警指标体系,该体系对于内部情况的指标设计参照了美国著名的CAMEL银行评级体系和中国银行业监督管理委员会颁布的《商业银行监管评级内部指引》,而外部威胁方面则从宏观经济环境、金融环境和国际收支环境三个角度设立相关指标。在指标体系的基础上构建了基于支持向量机方法的商业银行危机预警模型。为提高模型的精确度,在实证中考虑采用多种方法优化参数,并且最后得到了较高的分类准确率。对比实证结果,遗传算法在优化参数方面更具优势。采用遗传算法优化的支持向量机得到了高达96.6667%的分类准确率,该结果也同样验证了指标体系的有效性以及支持向量机方法的适用性。本文的特色与创新之处一是构建了支持向量机模型,验证了其在预警中国商业银行危机的有效性,并且检验遗传算法在其参数改进方面的优越性,解决了中国商业银行业样本有限、经营异质性强等特征下的危机预警难题；二是基于银行危机的诱发机理分析,构建了反映商业银行危机产生的内在根源、外来威胁两方面特征的预警指标体系；三是将民间融资规模、房地产和股市泡沫情况对银行危机的影响纳入研究中,弥补了现有研究只考虑社会信贷总额而忽视民间借贷与股市对银行存款的替代作用对银行影响的不足。
[Abstract]:Crisis early warning is an important proposition in the field of commercial bank management. Under the background of financial crisis in 2008, how to identify the characteristics of banking crisis and reveal the mechanism of bank crisis. Therefore, the construction of an accurate and effective early warning model has been paid more and more attention by the theorists and the practitioners. This paper begins with the definition of the banking crisis, the causes of the crisis, the indicators and methods of crisis warning, and systematically reviews the relevant literature on the banking crisis early warning research at home and abroad. Furthermore, the paper constructs the theoretical framework of this paper-the inducement analysis of the crisis of Chinese commercial banks and the early warning model based on support vector machine. On this basis, the internal causes and external threats of the crisis of commercial banks are considered. This paper analyzes the inducing mechanism of the crisis of Chinese commercial banks and establishes the early warning index system. The index design of this system for internal situation is based on the famous CAMEL bank rating system in the United States and the Internal guidance for Commercial Bank Supervision rating issued by the China Banking Regulatory Commission. External threats, on the other hand, come from the macroeconomic environment. In order to improve the accuracy of the model, the financial environment and the balance of payments environment establish the relevant indicators. On the basis of the index system, a commercial bank crisis warning model based on support vector machine (SVM) method is constructed in order to improve the accuracy of the model. In the empirical analysis, we consider using various methods to optimize the parameters, and finally get a higher classification accuracy. Compared with the empirical results. Genetic algorithm has more advantages in optimizing parameters. The support vector machine optimized by genetic algorithm has a classification accuracy of up to 96.6667%. The results also verify the effectiveness of the indicator system and the applicability of the support vector machine (SVM) method. The first feature and innovation of this paper is to build a support vector machine model to verify its effectiveness in early warning of the crisis of Chinese commercial banks and to test the superiority of genetic algorithm in improving its parameters. It solves the problem of crisis warning under the characteristics of limited sample and strong heterogeneity of management in China's commercial banking industry. Second, based on the analysis of the induced mechanism of the banking crisis, the paper constructs an early warning index system which reflects the internal root of the commercial bank crisis and the characteristics of the external threat. Third, the private financing scale, real estate and stock market bubble impact on the banking crisis into the study. It makes up for the deficiency of the existing research which only considers the total amount of social credit and neglects the substitution of private lending and stock market to the bank deposit.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F832.33;F224

【参考文献】