台灣股市當日沖銷、波動度、流動性關聯性分析

The Relationships Among Day-Trading, Volatility, and Liquidity in Taiwan Stock Market

鍾熾昌(CHIH-CHANG CHUNG)

指導教授 ： 林蒼祥

淡江大學/商管學院/財務金融學系碩士班/碩士(2013年)

https://doi.org/10.6846/TKU.2013.01027

摘要

本研究從時間序列的角度出發，利用高頻率的日內資料，接著再使用橫斷面分析方式去檢驗出台灣股市當日沖銷交易、市場波動度及流動性此三個變數在電子、金融及傳產類股中，彼此互動、影響的關係，當中並透過門檻向量自我迴歸模型 (Threshold Vector Autoregressive Model；TVAR )加以分析與討論，本研究資料以台灣證券交易所 2007 年 1 月 2 日至 2008 年 9 月 30 日，共計 451 個交易日之研究樣本，並進一步分析於此三大類股中在門檻效果下的關係。
根據本研究實證結果發現，於電子及傳產類股中，當於體制一(上)時，當日沖銷交易活動與股市波動度存在顯著正相關之影響力，即表示此二類股當波動度位於門檻值上時，愈多的當日沖銷交易會導致股市產生更多的波動，且波動亦與當日沖銷交易活動間存在正向回饋效果，顯示當沖者偏好高波動度的標的股票，當日沖銷交易對股市流動性之影響則有顯著正相關，即表示此二類股當波動度位於門檻值上時，愈多的當日沖銷交易導致其買賣價差縮小，提升市場之流動性，顯示當沖者亦會選擇流動性較佳的股票來進行交易；金融類股於體制一(上)時，當日沖銷交易是單向領先及預測市場波動度與流動性的，這可能是由於台灣股市金融標的股票之組成特性所導致，因其股性屬於穩定較不活潑型，故股價震盪幅度亦較小，而且根據高風險高報酬的財務觀點，金融類股較不適合當日沖銷交易此種投機意味濃厚之交易行為模式，因此較不易吸引當沖者的進場交易。綜合來說，當日沖銷交易、市場波動度及流動性，於體制一(上)之相互影響效果以電子類股最為顯著，其次為傳產類股，金融類股之表現則較不強烈。
於體制二(下)發現不論是電子、金融或傳產類股皆為低關聯性，顯示其變數間彼此是不具有領先及預測的關係存在的，故當波動度位於門檻值下時，於電子、金融與傳產類股來說，由於其價格的波動太低，以至於無法滿足當沖者的高風險高報酬期待，進而減少交易。
實證結果顯示，台灣股市之當日沖銷交易、波動度與流動性之間確實存在門檻效果，其影響程度則隨著各類股性質之不同而隨之改變。

關鍵字

當日沖銷交易 ； 波動度 ； 流動性 ； 門檻模型

並列摘要

The research employs high frequency of intraday data to examine relationships among day-trading, volatility and liquidity with electronic, financial and traditional industrial listed companies. The methodology used Threshold Vector Autoregressive Model, TVAR . The data of the research employed is from January 2,2007 to September 30,2008 in Taiwan Stock Exchange.
The empirical results reveal that under regime 1,day-trading and volatility exhibit highly positive correlation between electronic and traditional industrial listed companies, indicating that the higher day-trading frequency, the higher the degree of volatility in the stock market when the degree of volatility is above the threshold. Meanwhile, day-trading has high positive correlation with liquidity of the stock market if day-traders prefer high volatility stocks. In other words, the day-traders would rather trade the stock with higher degree of liquidity, since when the degree of volatility of electronic and traditional industrial listed companies is above threshold, the more day-trading frequency would cause bid-ask spread shrinking and enhance the liquidity of the stock market. the day-trading of financial listed companies exhibit one-way leading effect to predict both volatility and liquidity. These phenomenon could possibly be caused by the characteristics of financial listed companies, e.g. stability and less bouncing in price. Therefore, financial listed companies are less attractive to day-trader since day-trading usually stands for a risky strategy. In summary, electronic listed companies have the highest relationships among day-trading, volatility and liquidity, the next is traditional industrial listed companies. Financial listed companies have the least relationships among day-trading, volatility and liquidity.
The empirical results also reveal that under regime 2, the variables of electronic, financial or traditional industrial listed companies are neither leading nor predicting each other. Therefore, when the degree of volatility is under the threshold, for the financial, electronic, and traditional industrial listed companies, day-trader are not interested in trading since low volatility of price hardly satisfies their expectation to meet with higher risk premium.
The empirical results reveal that certain threshold effect do exist among day-trading, volatility and liquidity in Taiwan Stock Market. But the degree of influence will depend on the characteristics of each stock.

並列關鍵字

Day-Trading ； Volatility ； Liquidity ； Threshold model

參考文獻

3. Hu Shing-Yang and Chun-Kuei Hsieh,2010, “What Kind of Trading Drives Return Autocorrelation? ” Journal of Financial Studies, Vol.18 No.2,pp.65-98.

4. Tsay Ruey S,1989, “ Testing and Modeling Threshold Autoregressive Processes” Journal of the American Statistical Association, Vol. 84, No. 405, pp.231-240.

5. Atanasova Christina,2003, “Credit Market Imperfections and Business Cycle Dynamics: A Nonlinear Approach” Studies in Nonlinear Dynamics & Econometrics, Vol. 7, Issue 4, pp.1558-3708.

6. Bruce E. Hansen,1999, “Threshold Effects in Non-Dynamic Panels: Estimation, Testing, and Inference” Journal of Econometrics, Vol 93, Issue 2, pp 345–368.

7. Anderson, T.G., Bollerslev, T., Diebold, F.X. and Labys, P, 2001, “The Distribution of Realized Exchange Rate Volatility” Journal of the American Statistical American, Vol. 96, pp. 42-55.