金融危机的预警机制研究

发布时间：2018-01-22 22:35

本文关键词： 金融危机预警指标因子分析 BP神经网络　出处：《浙江大学》2014年硕士论文　论文类型：学位论文

【摘要】：在世界经济发展史上,曾爆发了多次金融危机,金融危机一旦爆发对世界范围内的经济市场都将产生不可估量的破坏,金融危机是世界经济市场发展史上极具破坏力的不定时炸弹。本文探寻预警金融危机的典型表现因子和数据,以提前预测金融危机为目的,构建了危机指标体系,探索了因子分析法和BP神经网络预警方法两种预警方法,并进行了实证研究。通过对国内外的文献研究对比分析,梳理了金融危机理论和金融危机预警方法及相关模型。对20世纪30年代的经济大萧条、1997年的亚洲金融危机和2007年的美国次贷危机进行对比研究,重点分析了危机爆发的原因,总结它们的表现形式,特别是相关经济指标的变化特征,为提取和识别金融危机的预警指标提供事实的基础。接着构建了金融危机的预警指标体系并进行了初步的数据处理,即分析选取了15个预警指标,为各指标确定了安全、基本安全、警惕、危险四个等级并确定了相关指标界限,查询了历年的统计数据,获取和计算了这些指标的数值并进行了归一化处理。之后阐述了因子分析方法的基本原理和操作流程,采用SPSS21.0软件对所获得的15个典型指标的统计数据进行了分析,采用主成分法提取了4个公共因子并按权系数计算了总因子,计算了因子得分系数,根据该系数计算了历年各因子的得分,并对各因子反映的经济状况进行了研究,结果表明,通过对因子得分的分析可以直观解析当前经济状况或者预警情况,为后期的经济政策的制定提供参考。最后在介绍BP神经网络的作用、模型特点和应用流程之后,基于MATLAB工具箱函数进行了BP神经网络构建,并选择了合适的传递函数和训练函数进行网络训练得出预测模型,验证了预测网络的泛化能力有效性后,对下一年的经济预警程度进行了预测,最后对预测结果进行了评析。结果表明,本文构建的预测方法,能有效预测下一年的经济情况。
[Abstract]:In the history of world economic development, there have been many financial crises. Once the financial crisis breaks out, it will cause incalculable damage to the world economic market. Financial crisis is a destructive time bomb in the history of world economic market. This paper explores the typical performance factors and data of early warning financial crisis in order to predict the financial crisis in advance. This paper constructs a crisis index system, explores two early warning methods, factor analysis and BP neural network, and makes an empirical study. Through the comparative analysis of domestic and foreign literature, this paper combs the theory of financial crisis, financial crisis warning methods and related models. On 1930s, the Great Depression. The Asian financial crisis in 1997 and the subprime mortgage crisis in the United States in 2007 were compared. The causes of the crisis were analyzed, and their forms of expression, especially the change characteristics of related economic indicators were summarized. For the extraction and identification of financial crisis warning indicators to provide a factual basis. Then it constructs the early warning index system of financial crisis and carries on the preliminary data processing, that is, analyzes and selects 15 early warning indicators, determines the security, the basic security, the vigilance for each index. The four grades of risk and the limits of related indexes are determined, the statistical data of the past years are queried, the values of these indexes are obtained and calculated, and the normalization processing is carried out. After that, the basic principle and operation flow of the factor analysis method are described, and the statistical data of the 15 typical indexes obtained are analyzed by SPSS21.0 software. Four common factors were extracted by principal component method, the total factors were calculated according to the weight coefficient, the factor score coefficients were calculated, and the scores of each factor over the years were calculated according to this coefficient. The results show that the analysis of factor scores can directly analyze the current economic situation or early warning situation, and provide a reference for the later formulation of economic policy. Finally, after introducing the function, model characteristics and application flow of BP neural network, the BP neural network is constructed based on MATLAB toolbox function. After selecting the appropriate transfer function and training function to train the network to get the prediction model, the validity of the generalization ability of the prediction network is verified, and the economic early warning degree of the next year is predicted. Finally, the prediction results are evaluated, and the results show that the prediction method can effectively predict the economic situation in the next year.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP18;F831.59

【参考文献】