Measuring And Analysing China's Business Cycle

Whether a country's macroeconomic is successful or not to a large extent depends on the judgment of the existing economics situation, in the function of macroeconomics, people think economic fluctuations as a whole mainly through the study of a series of economic variables. They also put forward the concept of'business cycle'to reflect the continuous ascent and descent stage of the regularity in the economy. Accordingly, the economic cycle is essentially a psychological feeling after people inspect the movement of a series of economic variables and between variables.In the long-run practice, people face two main problems in the process of measuring business cycle. First of all, there are so many indicators to reflect a country's economic situation and many differences exist in the business cycles, which makes it difficult to master the specific conditions of the economic cycle fully. Secondly, not all the indicators decline or rise at the same time when the economy declines from the top of prosperity or recovers from the bottom of depression, which makes the judgment toward the stage of the economic cycle inconsistent.The research of the first problem is based on the idea that the process of indicators fluctuations contains both long-run trends and short-run fluctuations. People propose a lot of methods to decompose indicators into trend components and cycle components, which range from moving average method, time trend model to later stochastic trend model, nonparametric trend model. Nowadays the latest filtering method is becoming mainstream. On the basis of traditional AR model and Fourier analysis, people take the nonlinear into account in the course of establishing models with cycle components. The analysis of indicators'cyclical fluctuations reach a deeper level by using wavelet analysis tools meanwhile. The research of the second problem is based on the fact that there exist comovement between a large number of economic indicators. People develop models of comovement between these indicators, and composite the business cycle indexes, from the Harvard index, diffusion index to composite index, SW index. Now the latest MS-SW index (that join the Markov switching)is being concerned by many scholars. People also considered the leading and lag relations of the movements between economic variables, introduce correlation analysis of the time difference methods and methods to inspect dynamically the leading and lag relations of the movements among these variables. They constitute the economic cycle index together with the comovement model, which improve the accuracy of choosing indicators.This paper drawn on the latest international research results and practical experience, systematically organizes and summarizes the cycle fluctuations'ways ,features, measurement and analysis to individual economic variables and a group of economic variables .It makes empirical analysis and contrast to the latest theories, models and methods combined with macroeconomic variables of our country, and tests its applicability in researching our country's economic cycle . This paper is structured as follows:Chapter One sums up the meaning and some stylized facts of business cycle, introduces some theory of explanations about business cycle, collates the course of measuring the business cycle and reviews briefly the newest developments of the business cycle.Chapter Two collates and summarizes systematically the literature of the existing domestic and international economic cycle measuring theory and model, and sums up specially for the nonlinear and synchronization issues, and reviews these documents simply.Chapter Three analyzes the cycle fluctuation of individual economic variable. Firstly, it introduces the study on the model decomposition methods and compares them. Secondly, it introduces the method of modeling the cycle fluctuation of individual economic cycle. Finally, it makes a empirical analysis on the composite index of the official announcement.Chapter Four summarizes the meaning of the potential output, the economics perspective of various schools and estimation methods and estimates our potential outputs quantitatively by using several methods that domestic documents have used. At last it compares the results of these methods based on the Fourier analysis and wavelet analysis tools.Chapter Five mainly analyzes the comovement and synchronization of economic variables. Firstly, it proposes several levels of comovement and synchronization of economic variables. Secondly, characterizing them through quantitative model and proposing testing methods or metrics. Finally, it also makes a empirical analysis on comovement and synchronization of China's key macroeconomic variables.Chapter Six is the extension of Chapter Five, it improves the traditional cyclical index method from the perspective of the comovement and synchronization, builds a dynamic factor model with a Markov transfer and a bandpass model, constructs MS-SW composite index and the bandpass composite index of our countries. By studying the business cycle measurement and analysis, this paper obtains the following results:1. It sums up and collates systematically the main economic variables measurement cycle fluctuation methods in the study on economic cycle, and discusses the latest filtering method emphatically. Three types of band-pass filter which is formed by filter have been compared. It's establishes that Butterworth band pass filter is the closest to the ideal band-pass filtering, ABP (12)is worse than it, and HP band-pass filter is the worst. But on the filtering process, they need to consume a certain amount of sample data, ABP and Butterworth both consume 12, HP only consumes four. From the filter parameters, ABP and Butterworth are both able to adjust parameters to consume more sample data and increase the effect of band-pass filter. But the effect of HP have no range to increase. Therefore, Butterworth is the optimal choice when do the indicators of economic trends and the cycle of decomposition if the sample data are long enough.2. Through the modeling analysis based on consistent composite index, we have summed up the characters of the economic cycle since 1991. We know that there are 44 and 18 months'cycle fluctuations in macroeconomic. Some regular features exist in our economic that expansion period is longer than depression period in prosperity, while in recession depression period is longer than expansion period.3. Using various methods appear at home and abroad, we estimate the potential output in China, and find that the results of all kinds of methods are similar. By analyzing through frequency domain, this paper found that there exist the short-cycle from 5 to 12 quarters, mid-and-short cycle from 13 to 24 quarters and mid-and-long cycle from 25 to 48 quarters in the business cycle in our country. Moreover, there exist the relation superimposed on each other in China's business cycle. When analyzing the mid-and-long cycle, the consequence of MV method is the best one; when analyzing the mid-and-short cycle, UC is the best one; when analyzing the short-cycle, HP is the best one ,MV is the worst one.4. By comovement this paper analyzing the consumable and CPI from January 1991 to February 2007 in four levels, we indicate that from November 1995, the common cycle feature starts to appear in the two indicators. The length of the identifiable common cycle is 39 months; from October 1995, the first order common dependent feature starts to appear in the two indicators. The dependent structure can be described by a MA(5). Common factor expressing comovement was analyzed by using model, the length of identifiable cycle is also 39 months.5. Analyzing industrial output and consumption indicators from January 1991 to December 2006 in the previous three synchronization levels of our country ,we conclude that the consequence of the correlation analysis of the time difference methods based on dual spectral density in the frequency domain is far better than that in the time domain, fluctuation of the two indicators with cycle length of seven months can reach maximum relevance of 0.69, which is better than the consequence of 0.42 in the time domain. In the bands with cycle length more than 15, industry output precedes consumption, and with the increasing of the cycle length, the phase shift increase constantly. At the maximum of the cycle, industry output and consumption are almost reverse phase. However, in the frequency bands from 0.08 to 0.18, the phase shift between fluctuations of two indicators are all low, and can be considered as synchronization. Inspecting the synchronization of two indicators dynamically, we conclude that in the steady macroeconomic environment, two indicators'synchronization are constantly increasing.6. Analyzing indicators in different groups through establishing the model of dynamic factors with Markov switching, we indicate that their consequences are consistent. Because of the consideration of nonlinear economic cycle, composite index constructed by the model is more effective than traditional one in describing business. The minor different in the different groups reflect the subjective for the indexes selecting and characteristic differences among indexes. By establishing frequency band model, it improves the accuracy of choosing indicators, and constructs separately long, mid-short cyclical index, and gives consistent measurement of business cycle for cycle fluctuations of different length.As a lot of new the theory and method measuring business cycle, the study of this paper and model experiment find some features of our country's business cycle and get some useful conclusions. As complement and reference of theory and method for measuring in our country's business cycle, all of these conclusions have important theoretical and practical significance.