探討國際股票市場是否具有效率性-門檻模型實證分析

An Exploration of the Efficiency in the International Stock Market-Based on the Empirical Analysis with the Threshold Model

洪怡琳(Yi-Lin,Hung)

指導教授 ： 陳俊志

國立臺北大學/社會科學學院/經濟學系/碩士(2012年)

摘要

當投資者決策股票是否存有可套利空間時，除了會觀察前一期股票價格，判斷未來股票價格的漲幅情勢，然而各個國家對於金融法規的制定不甚相同，外國人投資本國股票市場的進入障礙、交易手續費、漲跌幅限制，均是造成交易成本存在的原因，當交易成本存在時，各國股票市場從事交易，投資者是否會因交易成本存在，將其納入買賣股票時考量因素之一，造成存在不確定性而導致股票指數調整過程呈現非線性走勢，因此本文模型使用Obstfeld and Taylor (1997)提出之Band-TAR門檻模型。

本文實證結果發現，各國股票指數存在非收斂區間的門檻值十分顯著，除了上海、德國以外，韓國、香港、泰國、臺灣、英國、新加坡、日本、美國、深圳，均同時存在均數回復現象以及非收斂區間，落後一期的自我回歸模型實證結果也顯示出股票市場具有顯著的均數回復收斂情況，本文採用Harvey et al.(1997) 提出的Modified Diebold-Mariano檢定評估預測表現，選用2006年至2010年底為樣本外預測期間，從事四期向前預測，以落後一期自我迴歸模型及Random Walk模型與Band-TAR模型相比較，結果顯示Band-TAR模型在預測韓國、香港、臺灣、英國、新加坡、美國、深圳較落後一階自我迴歸模型捕捉到較多均數回復現象，但以Random Walk模型預測評估之表現並未顯著劣於Band-TAR模型。

關鍵字

門檻模型,均數回復

並列摘要

The investors will observe a previous stock price to determine whether arbitrage exists in the stock markets. However, the financial regulations in various countries are different. When making foreign investments, investors usually face the target country's stock market barriers such as transaction fees or price limits. Those are the reasons for the presence of transaction costs. Owing to the transaction costs, whether the investors take the costs into the consideration of the stock trading results is uncertain. It will cause that the stock index in the adjustment process to demonstrate a nonlinear trend. Therefore, we use the Obstfeld and Taylor (1997) Band-TAR model.

The empirical results indicate that the stock indexes in the non-convergence interval of the threshold are very significant, except for those indexes in Shanghai (China) and Germany. The rest of the countries, such as Korea, Hong Kong, Thailand, Taiwan, United Kingdom, Singapore, Japan, the United States and Shenzhen, show both the mean reversion and the non-convergence interval. In order to compare the linear and nonlinear models to see which one fits the adjustment process more, this paper uses the Harvey et al. (1997) Modified Diebold-Mariano test to assess the forecast performance. Then, we chose the period from the beginning of 2006 to the end of 2010 as the out-of-sample period and engaged in the four forward predictions and compared them with those in the previous event. Eventually, the results showed that the Band-TAR model, when used to predict the situations in South Korea, Hong Kong, Taiwan, United Kingdom , Singapore, United States and Shenzhen (China), demonstrated a more significant mean reversion phenomenon than the First-order Autoregressive model. However, the performance of the Random Walk model was not significantly worse than that of the Band-TAR model.

並列關鍵字

threshold model, mean reversion

參考文獻

﹝3﹞ 劉欣姿(1997)，以Band-TAR模型探討股票指數是否存在均數回復現象，國立台北大學經濟研究所碩士論文。

﹝4﹞ Amihud, Y., &Mendelson, H. (1986). Asset Pricing And The Bid-Ask Spread. Journal of Financial Economics, 17(2), 223-249.

﹝5﹞ Bai, J., &Ng, S. (2004). A Panic Attack On Unit Roots And Cointegration. Econometrica, 72(4), 1127-1177

﹝6﹞ Balvers, R., Wu, Y., & Gilliland, E. (2000). Mean Reversion across National Stock Markets and Parametric Contrarian Investment Strategies. The Journal of Finance, 55(2), 745-772.

﹝8﹞ Coakley, J., & Ana-Marı’A Fuertes (2001). Border Costs And Real Exchange Rate Dynamics In Europe. The Journal Of Policy Modeling, Journal Of Policy Modeling 23 (2001), 669–676.