本論文資料採用台灣、韓國與日本三國各股每日收盤價，研究之前先對資料進行基本統計分析，並對資料進行單跟檢定，避免資料存在假性迴歸，研究結果顯示原始資料存在單根，但在經過一階差分之後資料不存在單根，表示報酬率不存在單根，因此本論文皆以報酬率進行分析。由於文獻中如Chiang and Zheng(2010)等人說明羊群行為存在非線性的特性，因此本論文利用CSAD模型去觀察羊群行為，研究分成兩個部份分別做分析，第一利用CSAD模型討論各國是否存在羊群效應，第二則進一步加入馬可夫狀態轉換模型，觀察三國是否如Nofsinger and Sias(1999)等人文獻中的論述存在高波動度下較有羊群效應貢獻，以及文獻中Chiang and Zheng(2010)等人認為羊群行為具有非線性與不對稱的特性。 This essay adopt daily stock price of Taiwan, Korea and Japan. Before the research, we make a statistical analysis and unit root test for the data, in case the data exist spurious regression. The result of this research shows that the original data exist unit root, but after first-difference, the data do not exist unit root anymore, so as the return. This essay has been analyzed based of return. According to the literature, for example, Chiang and Zheng (2010) explained that herding behavior exist the characteristic of non-linear, therefore, this essay use CSAD model to observe the herding behavior. This research divides into two parts and analyzes respectively. The first, we use CSAD model to discuss if above mentioned three countries exist herding behavior. Second, we further add in Markov Switching model to observe if these three countries have relatively more contribution of herding behavior under high volatility environment, as the literature of Nofsinger and Sias (1999); and also as the literature of Chiang and Zheng (2010), who consider that there exist characteristic of non-linear and asymmetric.
