Stock Market Major Industry Indexes Are Predictable in Taiwan

• Main Authors: Ta-Lung Ho, 何大龍
• Other Authors: Tsun-Siou Lee
• Format: Others
• Language: zh-TW
• Published: 1993
• Online Access: http://ndltd.ncl.edu.tw/handle/19759221155338715236

碩士 === 國立臺灣大學 === 財務金融學系 === 81 === Earl tests of market efficiency, a large body of empirical literature seemed to support the stock market prices followed the random walks. It implied that returns were unpredictable from past returns or past variables. But a great deal of research found that stock returns seem to characterized by positive autocorrelation over intervals under a year and by negative autocorrelation over longer intervals. Moreover, we find that Taiwan stock market exhibit stronger and more singnificant deviations from the random walk hypothesis than the U.S. stock market. I used first-order autoregression, variance-ratio test and Markov chain test simultaneously, to examize Taiwan stock market index for the 1977-1992 period and found that total index and food index rejected the random walks. By using ARIMA to fit them, I found that every index wasn''t the random walk process and had predictable component. Every fitted model could explain 99.9% of the variance of the industry index and its mean square error only was 0.0025.