Study Of China’s Economic Cycle Characteristics Based On Mixed Frequency Model

Researches on characteristics of the economic cycle are not only important contents for describing the statue of the economic operation, but also prerequisites and foundations for monitoring economic fluctuations and making policy regulations. With the continuous development of econometric models and computer techniques, the understanding of the characteristics of the economic cycle is deepening. Under the circumstances adapting to New Normal of China ’s economy, it is of great significance to depict economic cycle characteristics efficiently and accurately, by studying the objective laws and characteristics of economic operation from the perspective of mixed-frequency data. Therefore, the characteristics of economic cycle are set as the research object, and mix-frequency data models are used to capture the fluctuating of the economic cycle characteristics, then a lot of suggestions about China ’s economic regulations and policies are put forward. The main research contents are the following.First of all, starting from the defining some connotation of economic cycle, combing with different frequency data which adapt to study different parts of economic cycle monitoring, the periodic features and variant features are well summarized. Based on that, he three part s of economic cycle characteristics that are the volatility, asymmetry and co-movement, are set as the research object of this paper. Then the formation mechanism of economic cycle’s characteristics is analyzed from the three aspects of internal evolution, external correlation and policy impacts, which shows that there are differences when measuring economic cycle by using data of different frequencies. Therefore, the analysis path of economic cycle is formed based on the application of mix-frequency data models.Secondly, based on the connotation definition of economic cycle, some economic indicators are chosen to reflect the overall situation of economic operation, in order to measure and analyze the economic cycle. According to the selecting criteria and the purpose of economic regulation, the gross domestic product, fixed asset investment, total retail sales of social consumer goods, t otal social power generation, as well as total export-import volume are chosen. Then, the common volatility factors of all indicators which are extracted by using the dynamic factor model, are reflecting the measuring results of China’s economic cycle, and based on these results the statues of economic cycle are recognized, as well as comparison analysis is completed. The results show that: for the one, the measurements of economic cycle measured by the dynamic factor model are stable, and can well reflect the common cyclical fluctuations of economic operation. For the other, the "valley- valley" method can recognize two complete economic cycles after 1992, and China has entered a stable normal statue since the third quarter of 2012.Thirdly, the Markov-Switching Mixed Data Sampling Model(MS-MIDAS) is used to analyze the fluctuated characteristics of China ’s economic cycle. Based on the introduction of different mix-frequency data models and their estimation methods, the mix-frequency data models for monitoring the fluctuations of China ’s economic cycle are constructed, and the real-time mix-frequency monitoring is carried out aiming at the fluctuations of economic cycle, by applying the indicator that is the growth rate of monetary supply into the model. According to comparing the fitting results of different regimes as well as forecast error of different models ’ forms, the mixed data sampling model with three regimes can help supervise the economic cycle to the best extend. Based on the selected model, the results of empirical study show that: China’s economic cycle fluctuations showed an obvious characteristic of three reg imes, namely growing with low speed, appropriate speed and rapid growth respectively; at the same time, changes of the three phases presented asymmetric characteristics in fluctuation amplitude and durability; Considering the predication ability, mixed-frequency data models perform better than same-frequency data model overall; mixed-frequency data model can timely capture fluctuations in economic cycles as it has better forecast accuracy and a certain sensitivity to trend of economic cycles.Fourthly, the Smooth-Transfer Mixed Data Sampling Model(ST-MIDAS) is used to study the asymmetrical characteristics of China ’s economic cycle. Based on the causes analysis of structural change, high frequency monthly data variables reflecting economic growth and quarterly indicators reflecting the economic fluctuations are chosen, in order to test structural change of economic cycle. The results of empirical studies show that: by using Z-A breakpoint root test, the structural change of economic cycle is confirmed; high-frequency data exert circulated influence on the asymmetrical characteristics of economic cycle, which make impacts that are different both the intensity of influence and time-lag influence, on economic cycle with different statues; the cyclic action of th e two extreme value obtained by mix-frequency data models can cover the results obtained by baseline regression models, so mix-frequency data models have better forecast accuracy than same-frequency data models.Fifthly, the Dynamic-Condition-Correlated Mixed Data Sampling Model(DCC-MIDAS) is used to analyze co-movement features between China’s economic cycles and financial variables. Based on analyzing the correlation and infectiousness between economic cycles and financial fluctuations, the mix-frequency data model which depicts the dynamic relationship between the two aspects is constructed to do empirical study of co-movement characteristics of China’s economic cycles. The results show that: there exists the co-movement characteristics of correlation and infectiousness between China’s economic cycles and financial variables; short-term impacts can infect the long-term trend of correlation, and the changing trend of influence coefficient has a time lag effect; through accumulation, short-term impacts can convert into long-term changing trend, which can better explain the long-term effect of dynamic correlation. For mix-frequency data models can capture more information contained in both high-frequency data and low-frequency data, it can explain the co-movement characteristics of economic cycle deeply.Sixthly, the Vector-Autoregressive Mixed-frequency Model(MF-MIDAS) is used to analyze macro-control policy of China’s economy. Aimed at performance characteristics of economic cycle, data about fiscal policy and monetary policy are chosen to analyze the impact effect of economic regulation on economic growth. The following conclusions are drawn: the influence that fiscal policy makes on economic growth is short-term and not significant; the influence that mone tary policy makes on economic growth is long-term and has a time-lag; the impose response obtained by VAR-MIDAS is more concentrated and compact, which means the estimation method is more effective than tradition method. For mix-frequency data models use more complete information of data, which can more accurately master the change of relationship between economic cycle features and policy impacts, it provides a more reliable method for policy study, and it is suitable for policy stimulation.Finally, on the basis of policy research, the simulation of economic cycle operation under different policy scenarios is conducted. The results show that there is difference between fiscal policy and monetary policy on economic cycle adjustment. The impact of fiscal policy on the economic cycle has the time lag effect, while the impact of monetary policy on the economic cycle is more sensitive. At the same time, the policy risk should be focused in the process of policy adjustment.