基于多层因子模型的中国动态 金融状况指数构建及其应用

作者	文清
姓名汉语拼音	Wen Qing
学号	2021000003033
培养单位	兰州财经大学
电话	15037697896
电子邮件	wenqing0207@163.com
入学年份	2021-9
学位类别	专业硕士
培养级别	硕士研究生
一级学科名称	应用统计
学科代码	0252
第一导师姓名	司颖华
第一导师姓名汉语拼音	Si Yinghua
第一导师单位	兰州财经大学
第一导师职称	教授
题名	基于多层因子模型的中国动态金融状况指数构建及其应用
英文题名	The Construction of China's Dynamic Financial Condition Index Based on Multi-Layer Factor Modeling and Its Application
关键词	金融状况指数混频动态因子模型多层因子模型谱分析
外文关键词	Financial Condition Index ; Mixed Dynamic Factor Model ; Multi-layered Factor Model ; Spectrum Analysis
摘要	随着金融改革开放深入推进，金融冲击对实体经济的影响愈发明显，在当前经济发展的新阶段和关键转型期，应对金融体系自身风险，是当前金融工作的重中之重。因此，构建能够及时反映金融市场变动和监测宏观经济走势的中国动态金融状况指数，具有重要意义。首先，本文阐述了金融状况指数的理论基础并且对现有的文献进行了总结，其次，以因子增广的时变参数向量自回归(TVP-FAVAR)模型的基础上，基于多层因子模型对金融因子的提取做出改进；然后选取利率、信贷、货币、房价、股价、汇率和宏观经济七类指标，利用改进的 TVP-FAVAR 模型测度中国动态金融状况指数（HDFCI）；最后，通过趋势图、分周期格兰杰因果检验、频域分析法和马尔可夫区制转换模型(MS-VAR)模型对构建指数和宏观经济变量之间的有效性、周期关联性和非对称性进行分析。通过实证结果发现：第一，在计算七类因子权重时，各权重具有时变性，而且各权重占比不一。房价、信贷因子、货币因子所占的权重较大，是影响我国通货膨胀的第一、二、三大传导因素；第二，在 FCI 构建方法上，分析并对比两种构建方式结果发现，本文构建的动态金融状况指数是一个相对较好的金融预警指标；第三，中国动态金融状况指数能有效反映宏观经济变动，在长周期上HDFCI 与 CPI、GDP 的相关性最大达 0.7，HDFCI 领先 CPI、GDP 大概 6 个月，说明在长周期上 HDFCI 对宏观经济的相关性和领先作用都更强。第四，通过马尔可夫区制转换(MS-VAR)分析发现在不同的经济金融状态下，金融市场对宏观经济的冲击存在明显的非对称性，在金融经济风险时期金融市场冲击对宏观经济的影响更大。本文拓展了金融状况指数构建的方法研究，构建出符合我国现实情况和经济走势的金融状况指数，可为相关政策制定提供科学依据。
英文摘要	With the deepening of financial reform and opening-up, the impact of financial shocks on the real economy has become more and more obvious, and coping with the financial system's own risks is the top priority of the current financial work in the new stage of economic development and the critical transition period. Therefore, it is of great significance to construct China's dynamic financial condition index, which can reflect the changes in the financial market and monitor the macroeconomic trend in a timely manner. Firstly, this paper describes the theoretical basis of the financial condition index and summarizes the existing literature. Secondly, based on the time-varying parameter vector autoregression model with factor augmentation, the extraction of financial factors is improved based on a multilayer factor model; then, seven types of indexes, namely, interest rate, credit, currency, house price, stock price, exchange rate and macroeconomic indicators, are selected to measure China's dynamic financial condition index by using the improved TVP-FAVAR model. FAVAR model to measure China's dynamic financial condition index; finally, the validity, cyclical correlation and asymmetry between the constructed index and macroeconomic variables are analyzed by trend diagram, sub-period Granger causality test, frequency domain analysis method and Markov-Systems-Variable-Area-Regime Transformation model. Through the empirical results, it is found that, firstly, when calculating the weights of the seven types of factors, the weights are time-varying and the weights have different proportions. The house price, credit factor, and monetary factor account for a larger weight, which are the first, second, and third major transmission factors affecting inflation in China; second, in the FCI construction method, analyzing and comparing the results of the two construction methods, it is found that the Dynamic Financial Condition Index constructed in this paper is a relatively good financial early warning indicator; third, China's Dynamic Financial Condition Index ( can effectively reflect the macroeconomic changes in the long run The correlation between HDFCI and CPI and GDP is up to 0.7, and HDFCI is ahead of CPI and GDP by about 6 months, which indicates that HDFCI has a stronger correlation and leading role in macroeconomics in the long run. Fourthly, the Markov System of Regions Transformation analysis finds that under different economic and financial states, there is an obvious asymmetry of financial market shocks to the macroeconomy, and the financial market shocks have a greater impact on the macroeconomy in the period of financial and economic risks. This paper expands the methodological research on the construction of the financial condition index, and constructs a financial condition index that conforms to China's reality and economic trend, which can provide a scientific basis for relevant policy formulation.
学位类型	硕士
答辩日期	2024-05-25
学位授予地点	甘肃省兰州市
语种	中文
论文总页数	64
参考文献总数	63
馆藏号	0005634
保密级别	公开
中图分类号	C8/410
文献类型	学位论文
条目标识符	http://ir.lzufe.edu.cn/handle/39EH0E1M/37007
专题	统计与数据科学学院
推荐引用方式 GB/T 7714	文清. 基于多层因子模型的中国动态金融状况指数构建及其应用[D]. 甘肃省兰州市. 兰州财经大学,2024.

条目包含的文件		下载所有文件
文件名称/大小	文献类型	版本类型	开放类型	使用许可
2021000003033.pdf（2625KB）	学位论文		开放获取	CC BY-NC-SA	浏览下载