Research On Development Of Dynamic Factor Models And Its Applications

Dynamic factor model has attracted great attention and become an important macro econometric branch in the past ten years. The basic function of dynamic factor models is to extract common factor from high dimensional data. In the study of dynamic factor model, the new research hotspots is the development of the basic dynamic factor model and the recognition of the factor's economics meaning, the corresponding is TVP- FAVAR model and DHFM model respectively. From the overseas research, the two models both have powerful application value in the field of policy analysis. However, there are few literatures about these two kind of models in China. In this context, this paper will focus on the theory and innovative application of TVP- FAVAR model and DHFM model. The main conclusions of this paper is as follows:First, in TVP-FAVAR models, this paper extracts three common factors from 115 dimension Chinese macroeconomic information set, and establishes a time-varying parameter factor-augmented VAR model(TVP-FAVAR) to study the effect of Chinese currency liquidity from 1996 to 2014. The results show that the three macroeconomic factors can depict Chinese macroeconomic movements, and the effect of Chinese currency liquidity on output, inflation and other key indicators has changed over the years. We also have found that TVP-FAVAR model has strong ability to capture the point information, and it has excellent performance in depicting the time-varying characteristic of currency liquidity transmission mechanism.Second, in TVP-FAVAR models, this paper examines the degree of international co-movements in inflation rates with a dynamic hierarchical factor model that decomposes 54 national inflation rates into global, regional, and country idiosyncratic components. The research results show that the global and regional factors together explain 55.7% inflation variability, which indicates that each countries' inflation rates have strong co-movement. However, the importance of the global and regional components differs substantially across countries. Though a cross-sectional regression, we find that the development measures, openness, average inflation and central bank independence strongly explain the national variation in the relative importance of international influences. Finally, a subsample analysis reveals that the international co-movement of China's inflation has significantly enhanced in recent years, and the imported inflation occupy about 45% of domestic inflation.In short, based on the detailed interpretation of the two types of extension model of dynamic factor model, this paper confirms that the expansion of dynamic factor model methodology has a broad application prospect in empirical analysis. Also, the study reflects the close tracking of international cutting-edge and innovation application in this paper.