中国股市羊群效应的区制转移时变性研究

传统研究采用静态CCK模型检验股票市场的羊群效应,但无法描述羊群行为的动态变化以及市场可能受到的外部影响。本文基于中国股市日频交易数据,在静态CCK模型中引入参数的区制转移性质识别股市在不同状态间的转换,并分析中国股市羊群效应和交叉羊群效应的时变特征。研究表明,中国股市运行周期可被划分为两个区制,分别呈现低波动和高波动的行情特征;羊群效应的程度随区制转移而变化,具有区制依存性。其中,沪深股市在高(低)波动区制中,羊群效应更强(弱),相应区制持续时间较短(长);中国台湾股市仅在高波动区制中出现羊群效应,相应区制持续时间较短;中国香港股市无论在低波动区制或是高波动区制中,均不存在羊群效应。此外,沪深A股在低波动区制中对美国股市和中国香港股市存在交叉羊群效应。

A Study of the Time-Varying Characteristics of Herding Effects in China's Stock Market Based on a Regime-Switching Model

Since 1990, the sharp rise and fall of asset prices has been a major issue in China's stock market. Ren et al. (2019) point out that one important reason for the drastic fluctuation of China's stock market is an investor structure dominated by individual investors. Due to limited information and insufficient rationality in investment decision-making, individual investors blindly follow stock market trends, which aggravates the fluctuation. Some scholars have argued that market information asymmetry and the herding effect caused by stock market participants' irrational activity lead to speculative froth in the stock market. For example, Liu et al. (2014) have found that under severe herding conditions, large irrational fluctuations in stock prices can lead to serious bubbles and financial crises. Tao (2017) finds that the synchronicity of Chinese stock prices is much higher than that of developed stock markets in Western countries, and herding behavior causes stock prices to rise or fall together. Therefore, identifying and analyzing the characteristics of herding behavior in China has practical significance for monitoring the stock market and providing early warning of risk. Domestic studies mainly use a static model, neglecting the time-varying characteristic of herding behavior. A few studies have discussed the dynamic characteristics of herding behavior in the U.S. market (Bohl et al., 2016), but these conclusions may not be applicable to the Chinese stock market, given its differences from Western stock markets in terms of system and development level. Furthermore, China's stock market has a complex structure, comprising the Shanghai, Shenzhen, Hong Kong, and Taiwan markets. Herding behavior may vary due to the different systems and market environments of different sub-markets. To solve these problems, we use a Markov-Switching CCK model to analyze dynamic herding behavior and cross-herding behavior in China's segmented stock markets within a regime-changing environment.Our sample was drawn from all of China's segmented stock markets during the 1997-2019 period. The data were collected from the Wind, CSMAR, and DataStream databases. The following findings were obtained based on empirical analysis. The cycle of Chinese stock markets can be divided into two regimes characterized by high volatility and low volatility respectively, and the intensity of the herding effect varies with the transition between regimes. For the Shanghai and Shenzhen stock markets, the herding effect is relatively brief and intense in the high-volatility regime, while in the low-volatility regime, the herding effect is longer but relatively mild. For the Taiwan stock market, the herding effect is found only in the high-volatility regime, and lasts for a short time. For the Hong Kong stock market, the herding effect does not exist in either the low-volatility or high-volatility regime. In addition, the A-share markets herd around the U.S. and Hong Kong markets during the low-volatility regime.The contributions of this paper are as follows. First, considering the special structure of China's stock market, we discuss the herding effect under different systems and market environments, which provides complementary evidence for the time-varying herding effect of China's segmented stock markets under different regimes. Moreover, we contribute to the literature on financial risk contagion. Based on the perspective of cross-border financial linkage, we extend the basic CCK model to the cross-border financial context. By examining the time-varying characteristics of cross-herding, this paper highlights the herding path of risk contagion between stock markets, thus deepening the understanding of financial risk contagion and financial risk management in an open environment. Second, to obtain more rigorous conclusions, we perform necessary parametric tests in the empirical analysis. This can be regarded as a valid method for performing an availability analysis of a Markov switching CCK model, which provides valuable guidance for scholars to conduct more comprehensive, more scientific and more effective research on such a model.