The Trend And Cycle Of China’s Stock Market:Decomposition Method,Theory And Empirical Analysis

As a representative of emerging markets,China’s stock market is an important investment place in the country and even the world.In the past 30 years,China has maintained its position as the world’s second-largest economy.However,against the background of rapid economic development,China’s stock market has been in a state of instability for a long time and has experienced several sharp rises and falls.Scholars have tried to explain this anomaly in terms of policies,regulations and investor characteristics,but they generally only focus on stock price itself.There is little literature on in-depth analysis of the characteristics and factors of China’s stock market and its correlation with macroeconomic fundamentals from the perspective of longterm trends and short-term cycles.As a result,it is difficult to accurately determine whether China’s stock market has actually deviated from its fundamentals and the degree of deviation in the long-term.On the basis of summarizing the typical characteristics of China’s stock market,this paper proposed new methods to separate the trend and cycle of stock price,and then established a systematic analysis framework to deeply explore the causes of abnormal fluctuations in China’s stock market.At present,the modeling and trend-cycle decomposition methods of economic time series are abundant,but they are still limited to I(1)processes.However,in reality,most nonstationary series are not strictly I(1)processes.For example,asset price series with long memory such as house price and stock price are usually fractional difference stationary processes.How to decompose the trend and cycle of the fractional difference stationary process is worthy of further study.In this paper,considering the characteristics of long memory,fractional cointegration and structural changes in China’s stock market,we extended the traditional trend-cycle decomposition method and proposed some new univariate or multivariate models and theirs trend-cycle decomposition methods,including the ARFIMA-GARCH model with symmetric threshold and the FCVAR model with Markov regime switching.To analyze the long-term trend and its key factors of China’s stock market,we theoretically considered a simple pricing model that can reflect the relationship between stock price and the macro economy by maximizing the utility of investors in a singlesector production economy.It is derived from the pricing model that output-capital ratio,investment-capital ratio and productivity are the main fundamental factors of stock price.Empirically,we innovatively applied the MS-FCVAR model and its trend-cycle decomposition method,which can describe structural changes and fractional cointegration,to China’s stock market,and obtained two important findings.First,around the fourth quarter of 2007,the long-term cointegration relationship between variables has changed structurally.This is mainly reflected in the impact of the outputcapital ratio on the stock price turns from negative to positive.The main reason is that the driving force for economic growth has changed.Second,there is a common trend between stock price and productivity,but there is no common cycle between stock price and any other variables in the cointegration system.In other words,from the long-term trend,there is a very close correlation between stock price and the macro economy,and the most critical factor is productivity.From the short-term cycle,the correlation between stock price and the macro economy is very weak.To analyze the cyclical fluctuation and its key factors of China’s stock market,we mainly focused on irrational extrapolative expectations of investors.Theoretically,we proposed a productivity-based pricing model which features long-run productivity risks and irrational extrapolative expectations and thus successfully explained the puzzles of risk premium and large volatility in China’s stock price.Empirically,we used the simulated data of extrapolative expectations to build a SVAR model and then analyzed the short-term influencing mechanism between stock price and its factors.The variables include the cycle component of stock price,extrapolative expectations,PPI,investment,and gross fixed capital formation.The results show that investor’s extrapolative expectation has a significant effect on the cyclical component of stock price.The trend-cycle decomposition of stock price can also be applied to the trend prediction and turning point identification of stock market.In this paper,after decomposing the trend and cycle components of stock price based on an ARFIMA model,we used the wavelet leaders method to analyze the multifractal characteristics of the cycle component and further proposed two new indicators to detect market turning points.One is the third-order multifractal parameters of wavelet leaders based scaling exponent.The other is the discrepancies of scaling exponents of the wavelet leaders and the discrete wavelet transform.Empirically,applying them to the US and China stock markets,we found that both indicators have good performance in detecting critical turning points,as well as high-volatility and high-risk points.More importantly,through comparison tests with other identification indicator of existing literature,our indicators stand out with prominent detection performances.