東亞地區股票市場之聯動性分析

Abstract

摘要 本研究擬探討，在全球市場聯動性愈來愈增長的背景下，任何一個較為重要地區之股票市場受到波動或者突發性事件，都有可能波及或牽涉到其他國家與地區的股票金融市場。在本篇報告中，於探究東亞四個主要股市市場，分別為韓國的“KOSPI指數” ，日本的“日經指數” ，台灣的“台灣加權指數以及”香港的“恆生指數”此四個指數皆為該國股票市場中的重要指標，除了對東亞此四個地區的股票市場個別作探討更著重於股票市場之聯動性進行分析。報告中我們擷取2010年12月1日至2018年12月1日之股市數據，而我們採用的數據為日報酬，其特徵為擁有群集波動性及厚尾高峽峰，而本篇著重於了解東亞地區股市之聯動性，因此只要其中一個股市未開市，我們便將該筆(日)資料刪除，得出有效資料為1716筆，並提供基礎統計量及報酬率的時間序列圖，由資料顯示峰態係數皆大於常態分布(3)，意即屬於高峽峰的型態，並且為左偏分配。並且透過一些檢驗，發現此四個市場皆具有ARCH效果，並且不具常態分布以及呈現不對稱性，透過嘗試後發現EGARCH具偏Student-t誤差之模型對於此四個市場是最合適的模型。最後採用多元變異數異質性(DCC-GARCH)模型對此四個地區作聯動性分析，結果顯示此四個股市間具有波動性的聯繫，由相關係數圖可看出市場間相關性的起伏，藉由它的起伏我們查出了導致其相關性較低的緣由，從條件共變異數圖可得知兩兩市場中在哪些期間下變數的變化趨向為一致。最後觀察出各股票市場之間，存在有較高的市場聯動性。此外，我們通過政治、經濟、貿易等方面因素，對各股票市場中出現的較大同段波動之部分進行了分析與解釋。

Abstract The purpose of the study is to explore that under the background of increasing interconnectedness of global markets, any stock market in an important region would be affected by markets from other countries and regions due to fluctuations or unexpected events. In this report, we are order to explore four major indexes from different countries’stock markets in east Asia. They are South Korea's "KOSPI" index", Japan's "Nikkei" index, Taiwan's "Taiwan Weighted" index and Hong Kong's "Hang Seng" index. Every of the four index is an important indicator in the country's stock market, in addition to exploring the stock market in the four areas individually ,we plan to focus more energy on Analysis on the linkage of stock markets in these area. In the report,we collected the data consisted of daily closing values from December 1, 2010 to December 1, 2018, and then we got the daily stock-return series by taking the logarithmic difference of the daily stock-index times 100.That is characterized by volatility clustering and thick tail high peak.This paper focuses on the correlation of understanding the stock market in east Asia, so as long as one of the stock market is not open, then we will delete that day’s data, it is concluded that effective data for 1716, and provided the basic statistics and return on time sequence diagram.The data showed kurtosis coefficient is greater than the normal distribution , which belongs to high peak type, and distribution for left.Through some tests, it was found that all the four markets had ARCH effects , and there was no normal distribution or asymmetry. After trial, it was found that the EGARCH model with biased Student-t error was the most suitable model for the four markets.Finally we usd multivariate variance heterogeneity (DCC - GARCH) model for four areas to complete a correlation analysis, the result showed that the four has the volatility of the relationship between stock market.According the correlation coefficient graph we can see the correlated volatility among markets.From the conditional covariance graph, it can be known that the variation trend of the variables in two markets is the same during which period. Finally, we observed that there is a high market linkage among stock markets.In addition, through the political, economic, trade and other aspects of factors, the stock market appeared in the larger part of the same period of fluctuations are analyzed and explained.