动态因子模型的拓展研究及应用

发布时间：2018-05-17 02:10

本文选题：动态因子模型 + TVP-FAVAR模型　；参考：《华中科技大学》2016年硕士论文

【摘要】：通过从高维经济信息集中提取共同因子,以此识别驱动宏观经济系统变化的共同因素,动态因子模型(DFM)在过去十年引起了学者们极大的关注,已然发展为了宏观计量经济学的重要分支。在动态因子模型的理论研究方面,目前最新的研究热点为对基本动态因子模型的拓展和识别因子的经济学含义两个方向,相对应的分别是TVP-FAVAR模型和DHFM模型。从国外研究来看,这两类模型在宏观经济政策分析中均具有强大的应用价值。从国内看来,研究这两类模型的文献极少。在此背景下,本文深入研究了动态因子模型最新拓展形成的前沿方法:TVP-FAVAR模型和DHFM模型。本文的研究内容分为两个层次:一是模型理论的介绍,二是方法论成果的创新性应用。本文的主要结论如下:第一,在利用TVP-FAVAR模型对我国货币流动性效应分析的应用研究中,本文构建了115维宏观经济信息集并从中提取宏观共同因子,以此建立了因子增广的时变参数向量自回归模型(TVP—FAVAR),对中国1996—2014年间的货币流动性效应进行了研究。实证结果表明:3个宏观因子基本能刻画我国宏观经济的动态特征,中国的货币流动性对产出、通货膨胀等的影响机制在样本期内发生了显著的结构性变化,TVP—FAVAR模型对时点信息的捕捉能力较强,能够很好的刻画中国货币流动性传导机制中的时变特征。第二,在利用DHFM模型对通货膨胀的国际协动性研究的应用研究中,本文利用动态分层因子模型,将54个国家1980-2014年的CPI数据分解为全球通胀因子、区域通胀因子和国家特质成分,以探究各国通胀的协动性和波动来源。研究结果表明,全球通胀因子和区域通胀因子合计解释了各国通胀波动的55.7%,表明各国通胀具有较强的协动性。然而,全球因素和区域因素对不同国家通胀的影响存在显著的差异,通过截面回归发现,经济和金融发展水平、开放程度、平均通胀率以及央行独立性是造成这种国别差异的重要因素。最后,子样本分析显示,近年来中国通胀的国际协动性明显增强,输入型通胀大约占据国内通胀成分的45%。总之,本文在详细解读动态因子模型的两类拓展模型的基础上,证实了动态因子模型的拓展方法论在实证分析中具有广阔的应用前景,体现了本文对国际前沿的紧密跟踪和应用性创新。
[Abstract]:By extracting common factors from high-dimensional economic information to identify the common factors driving macroeconomic system change, dynamic factor model (DFM) has attracted great attention of scholars in the past decade. Has developed into an important branch of macroeconometrics. In the theoretical research of the dynamic factor model, the latest research focus is the expansion of the basic dynamic factor model and the economic meaning of the identification factor, corresponding to the TVP-FAVAR model and the DHFM model respectively. From the abroad research, the two models have strong application value in macroeconomic policy analysis. From the domestic point of view, there is very little literature on these two models. Under this background, this paper deeply studies the new frontier methods of dynamic factor model: TVP-FAVAR model and DHFM model. The content of this paper is divided into two levels: one is the introduction of model theory, the other is the innovative application of methodology. The main conclusions of this paper are as follows: first, in the application of TVP-FAVAR model to the analysis of monetary liquidity effect in China, this paper constructs a 115-dimensional macroeconomic information set and extracts macro-common factors from it. Based on this, a factor augmented time-varying parameter vector autoregressive model (TVP-FAVARA) is established to study the effect of monetary liquidity in China from 1996 to 2014. The empirical results show that the three macro factors can basically depict the dynamic characteristics of China's macro economy. The influence mechanism of inflation and so on has undergone significant structural changes in the sample period. TVP-FAVAR model has a strong ability to capture time point information and can well depict the time-varying characteristics of the monetary liquidity transmission mechanism in China. Secondly, in the application of DHFM model to the international cooperative study of inflation, this paper uses the dynamic stratified factor model to decompose the CPI data of 54 countries from 1980 to 2014 into global inflation factors, regional inflation factors and national characteristics. In order to explore the cooperativeness and volatility of inflation in various countries. The results show that the global inflation factor and regional inflation factor together explain the fluctuation of inflation in each country 55.7. it shows that the inflation of each country has a strong synergism. However, there are significant differences in the impact of global and regional factors on inflation in different countries. By cross-section regression, we find that the level of economic and financial development, the degree of openness, Average inflation and central bank independence are important factors contributing to such country-to-country differences. Finally, subsample analysis shows that China's inflation has increased significantly in recent years, with imported inflation accounting for about 45% of domestic inflation. In a word, based on the detailed interpretation of two kinds of extended models of dynamic factor model, this paper proves that the expansion methodology of dynamic factor model has a broad application prospect in empirical analysis. This paper reflects the international frontier of the close tracking and application of innovation.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：F224;F821

【参考文献】