节点文献

基于数据挖掘技术的VaR方法在信用风险度量中的应用

Application of Value-at-Risk Measures on Evaluation Credit Risk Based on Data Mining Technology

【作者】 朱莉

【导师】 匡建超;

【作者基本信息】 成都理工大学 , 管理科学与工程, 2004, 硕士

【摘要】 信用风险作为一种最重要的金融风险,在经济生活中受到越来越多的关注。对信用风险的研究,无论是在金融理论还是金融实践方面,都发生了巨大变化。目前,国际上对信用风险的度量正从定性研究为主向定量研究方面发展,这一发展趋势将更加强调利用信用风险模型来度量信用风险,进一步推动了信用风险管理技术的变革,信用风险的度量与防范已是摆在中国金融业面前的迫切课题。 近年来,随着我国市场经济的飞速发展和竞争的不断加剧,各个企业、金融等部门的领导人,愈加认识到仅靠传统的经验式决策是行不通的,但他们往往不是从数学模型出发而是从经营的需要出发提出要解决的问题。管理者对方法、模型缺乏全面了解,造成应用与理论方法的脱节;另一方面,计算机技术的不断进步,使计算机从单纯的科学计算,发展到被应用于社会生活的各个角落,利用先进的计算机技术来实现智能化的信用风险度量,是今后发展的一种必然趋势。 在对信用风险管理状况作了适当综述后,文章重点研究了在信用风险度量模型的构建中数据仓库和数据挖掘技术的运用,以及通过极值方法计算在险价值、度量信用风险这两个方面。较好地克服了传统信用风险度量模型中的缺陷,大大提高了系统的智能化程度。本文以VaR技术为核心,提供一套从数据库管理、IT系统监控和分析到信用风险控制和评估机制的全面解决的方案,在探索VaR风险度量模型在我国信用风险管理及实践方面做出了一定的尝试。最后,在对结论进行了数量分析的基础上,针对目前我国信用风险管理的状况,提出了一些意见和建议,为我国信用风险监管事业的发展提供了一定的理论依据。 本文在信用风险度量模型的系统构架研究方面作进行了有益的研究和探索,取得的成果如下: ①在对信用风险管理中涉及到的一些相关概念和内容进行了综合论述后,针对国内外信用风险管理现状进行了对比,然后分别从企业和银行两个角度阐明了目前我国在信用风险监管的方面还存在的一些不足之处; ②对几个常用的信用风险度量模型及其优缺点作了比较,揭示了它们在信用风险度量的实际操作中存在的问题; ③将数据仓库和数据挖掘技术运用到信用风险模型的构建中,进一步提高了整个系统的智能化程度,为信用风险的准确度量奠定了坚实的基础; ④采用目前极具代表性的VaR技术来对信用风险进行度量,并利用极值方法来计算在险价值,从而克服以往信用风险度量模型中可能存在的不足,在信用风险度量研究上迈出了重要一步; ⑤通过实证研究指出了目前VaR方法在我国应用的局限性和相应的对策,并就完善我国信用风险管理体系方面提出了一些建设性意见,为今后在信用风险度量领域的进一步研究提供了一些思路。 本文的根本出发点是结合经济生活中企业和金融部门可能出现的信用状况和问题,寻找一种行之有效的方法来更好地量化信用风险,以便达到对其进行评估和监控的目的。这样,中国的许多企业和金融部门才能在未来更加激烈的全球化市场竞争中占有一席之地,真正意义上地实现风险-受益的优化配比。

【Abstract】 As an important part of financial risks, credit risk has been paid more and more attentions in economic life. Research to credit risk changes greatly both in financial theory and practice. At present, evaluation of credit risk is shifting from qualitative analysis to quantitative analysis, this trend will emphasize on evaluating credit risk by mathematical models and speed up the revolution of credit risk management. So that, testing and supervising the credit risk is crucial to finance in China.Recent years, with high-speed development of market economy and deep competition in China, policymakers of enterprises and financial offices begin to realize that it’s not enough to make decisions by traditional experience. Poor understanding of mathematical methods and models has made theories disjointed with practices. On the other hands, with the renewing of computer science, it has been used widely. It’s a trend to evaluate credit risk intelligently by computer techniques. This thesis will undertake by follow researches and exploration about the construction of a computer assistant "Evaluating Credit Risk Model": (1) Comprehensively described some terms and contents in credit risk management, andcompared credit risk management situation in China to foreign institutions. Putforward certain defects in China from different views of enterprises and banksseparately. (2) Analyzed and compared several credit risk models in common use and their virtuesand weaknesses. (3) Applying the technique of Data Ware and Data Mining to the construction of creditrisk models to further improve the intellectualized degree of this model. (3) This article will use the typical VaR Measure into evaluation of credit risk andcalculate the Value-at-Risk by Extreme Value Method, overcoming the weakness oftraditional models.The starting point of this thesis is to find a powerful way to quantify credit risk combining with possible credit problems. This article emphasizes on the application of Data Ware and Data Mining techniques in the construction of credit risk model and evaluation of credit risk by Extreme Value Method. Effectively overcame defects in traditional models and greatly improved the intellectual degree of this model. At last, based on quantitative analysis, the author advanced some suggestions, which aimed at present credit risk management situation of our country, and provided certain theoretical basis for the career of credit risk supervision.

  • 【分类号】F224
  • 【被引频次】2
  • 【下载频次】751
节点文献中: 

本文链接的文献网络图示:

本文的引文网络