Trend-Cycle Decomposition And China’s Business Cycles

Business cycles which are the phenomenon of repeated expansions and contractions in market economy are the focus of macroeconomics. Since Keynes, each school of macroeconomics tries to create a new theory to explain the business cycles. How to evaluate these competing theories has become the main content of the macroeconomics research. Empirical studies based on statistical data are the main choice, and the premise of these empirical studies is to extract the cycle component from the data, which is the focus of the dissertation, trend-cycle decomposition methods. Since2004, domestic researches in this field are gradually increasing, but most of them are in the primary stage.According to the methodology, trend-cycle decomposition methods are divided into single variable decomposition and multivariate decomposition in the dissertation, and the former are further divided into decomposition based on probability theory and filtering methods. The structure of the dissertation is also arranged accordingly. The followings are the main content and innovation of the dissertation.For single variable decomposition, the innovative research of the dissertation is the methodological research of the sign problem of BN decomposition, which has not been solved yet. The dissertation establishes the quantitative relation between the sign problem of BN decomposition and the persistence measurement indicator. And the results show that when the VR indicator is greater than one, the time series has strong persistence and the sign problem will occur, otherwise does not. The dissertation finds the sign problem may also occur in the cycle of unobserved component model decomposition method, and there is also the similar quantitative relation. The above are the theoretical researches in the field of trend-cycle decomposition firstly.Based on the methodological research in single variable decomposition, focusing on the composed cycles, the dissertation proposes the standards of periodic, stability, timeliness and consistency to evaluate the decomposition methods. Periodic focuses on period of the cycle components. Stability focuses on the difference between the latest and the former cycle component after adding new observational data. Timeliness focuses on whether the last value of the cycle component is able to reflect the status of this variable accurately. Consistency focuses on the difference between the cycles coming from the data of different frequency. Focusing on Chinese real GDP, the research shows that CF filtering is relatively better. From the perspective of competitive research, the standards proposed by the dissertation are able to evaluate the decomposition methods well. The above researches reflect the innovative.In the field of trend-cycle decomposition, multivariate decomposition is the most frontier and most difficult research direction. Following the main papers, from the perspective of methodology, the dissertation expresses the development and the theory clearly. From the existing literatures, this is the first time in domestic research. Specifically, the multivariate trend-cycle decomposition is based on non-stationary time series theory, decomposes the variables after taking into account the interaction between trends and cycles. So, the dissertation studies the interaction between trends using the co-integration theory firstly, and then studies the interaction between cycles using the serial correlation common feature theory proposed by Engle and Kozicki (1993). On this basis of the above, the multivariable trend-cycle decomposition is achieved through VEC model constrained by co-integration and serial correlation common feature. On the basis of theoretical research in the multivariate trend-cycle decomposition, the dissertation is carried out applied research. The research on money and business cycles shows that the money factor is important in business cycles and is the leading indicator. The research on agricultural price and inflation shows that the codependence between the agricultural price and CPI is weak, and the fluctuations of agricultural price are the impact effects of rises and falls of the grain and pork prices to a large extent. The codependence between the monetary and CPI is strong, which reveals that the monetary policy target alternatively transforms between preventing inflation and promoting growth. The policy recommendations are on the period of the rising of the agricultural prices, to control inflation, the agriculture price should be suppressed firstly by supply side.In summary, the theoretical and methodological innovations in the dissertation can be summarized as the followings:the research of the sign problem of the BN decomposition and unobserved components model decomposition; the research of the standards to evaluate the single variable decomposition The innovation of application research is that using the latest multivariate trend-cycle decomposition method, from the important and not fully concerned perspective, the economic operation and inflation problem are researched, and the practically significant conclusions and applicable policy recommendations are achieved.