(3.231.29.122) 您好!臺灣時間:2021/02/25 16:48
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果

詳目顯示:::

我願授權國圖
: 
twitterline
研究生:王惠中
研究生(外文):Wang, Hui-Chung
論文名稱:臺灣股市弱式效率檢定與異質變異數模型之應用
論文名稱(外文):Tests of Market Efficiency and The Application of Conditional Hetero-scedasticity Models In Taiwan''s Stock Market
指導教授:劉聰衡劉聰衡引用關係
指導教授(外文):Liu, Tsung-Heng
學位類別:碩士
校院名稱:淡江大學
系所名稱:金融研究所
學門:商業及管理學門
學類:財務金融學類
論文種類:學術論文
論文出版年:1991
畢業學年度:79
語文別:中文
中文關鍵詞:變異數比率法迴歸斜率法
外文關鍵詞:Variance Ratio TestRegression Slope Test
相關次數:
  • 被引用被引用:4
  • 點閱點閱:138
  • 評分評分:系統版面圖檔系統版面圖檔系統版面圖檔系統版面圖檔系統版面圖檔
  • 下載下載:0
  • 收藏至我的研究室書目清單書目收藏:0
  本文的實證第一階段對台灣股市之加權股價指數,以均方庇連差異法(Mean Square Successive Difference Test),變異數比率法(Variance Ratio Test),迴歸斜率法(Regression Slope Test),檢驗市場是否為一隨機漫步假設。所獲之結論如下:
  1. 股價報酬隨所衡量的時間水平(time horizon)增加,負的相關愈明顯,表示股價在長期有回復均值現象。
  2. 股價報酬在短期有正的相關。
  3. 股價報酬序列應不是全然的隨機漫步,即視為一平穩隨機慢步(stationary random walk)序列。
  4. 股價中含有暫時性成份,這部份隱含報酬是可預測的。
  三個檢驗方法所獲結果都一致拒絕隨機漫步假說。因此,弱式效率在台灣股市應不成立。
  藉由上述的結果。適可引入第二部份的實證,因不管是短期或長期,股價報酬序列間都有相關性存在,所以對其配適─自我迴歸的模型,並引用異質變異數模型ARCH,GARCH,ARCH-M,CARCH-M。在所檢定三個子期問與母期間中,第二子期間(民國六十七年到七十四年)並不具有ARCH或CARCH效果,其它期間都有顯著的效果。
  以最大概似法的估計都能很快的收斂,所估得GARCH的參數和都接進一,表示變異性對報酬所造成的衝擊有很高的持續性。在ARCHH-M和GARCH-M應用模型中,估計出市場的相對風險規避係數,其值都比國外所估出的值要小,顯示台灣股市,風儉的價格比國外要低,是否台灣的股市預期報酬的考量因素中,風險不是一重要的變數?則需要更深一層的研究。
  另外在應用預測能力上,比較ARCH,CMRCH,和移動平均法對變異性的預測能力。ARCH,CARCH都明顯得比移動平均法要好.而CARCH的預測誤差又比ARCHH略小。
  This paper tests the random walk hypothesis of the Taiwan Stock Market, using the following three methods, (l) Mean Square Successive Difference test. (2) Variance Ratio test. (3) Regression Slope test. The results from this empirical work show that returns exhibit positive autocorrelation over short horizon and negative autocorrelation over long horizon. Stock prices also exhibit mean-reversion phenomenon and weekly return series is not a pure random but a stationary random walk. This means that returns are predictable.
  Return series also exhabit a significant first-order autocorrelation and a significant levels of second-moments dependence. A reasonable return generating process is empirically shown to be a first-order autoregressive process with conditionally heteroscedastic innovations. We employ ARCH, GARCH, ARCH-H, and GARCH-M to capture the dependence of squared residual series. The ARCH or GARCH effects are significant in periods of 1971-1977, 1986-1991, and 1971-1991. The ARCH-M or GARCH-H effects are insignificant in all periods except GARCH-H in the entire sample.
  Maximization of log-likelihood functions converges to the optimum rapidly in all cases. The sum of parameters in GARCH model closes to unit, indicating that volatility of stock market fluctuations persists for a long time. The investors'' coefficients of relative risk aversion in the Taiwan Stock Market is saaller than those of foreign stock markets. This indicates that price of risk in the Taiwan Stock Market is cheaper than that in foreign stock markets. However, we need further studies to investigate whether risk is an important factor determining expected return in the Taiwan Stock Market or not.
  The ARCH and GARCH models are applied to forecast weekly volatility of returns and compared to Moving Average method. Our result is similar to other studies that ARCH and CARCH are much better than Moving Average. At the mean time, the GARCH has the best fitness of volatility.
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top
系統版面圖檔 系統版面圖檔