| Summary: | 碩士 === 國立屏東科技大學 === 財務金融研究所 === 95 === In the finance area, the long term asset return behavior is an important topic. The so-called slow long memory or the long memory refers to the time series that the auto-correlation function dies away following the negative power index, instead of following the index ratio which dies away rapidly or linear pattern which dies away slowly. In other words, the historical events still have the influence in a longer term to the future. This also means the time series follows fractional Brownian motion.
This research aims to study the existence of long memory and volatilities clustering for Taiwan stock market. Four models are employed, namely ARIMA, ARIMA-GARCH, ARFIMA, and ARFIMA-GARCH in this paper. The empirical results show that the return of Finance, Plastics and Chemicals and Electric Machinery stock price index exist long memory. It shows that during the study period, the stock market is not efficient.
The long memory in stock returns implies that the previous volatilities have impacts in the future stock returns. This also means that the future stock returns can be predicted using past returns. Thus the empirical results provide important implications to return predict and investment risk control.
|
|---|
