| Summary: | 碩士 === 淡江大學 === 財務金融學系碩士班 === 93 === This paper selects data from JAN.4th, 2000 to AUG.29th, 2003, total 881 daily data. Employing the time series techniques, investigating S&P 500 stock index close price、Taiwan futures or spot stock index close price which have significant leading affect for Taiwan stock index open price. At the same time , investigating NASDAQ stock index close price、Taiwan election futures or spot stock index close price which have significant leading affect for Taiwan election stock index open price. Employing study methods including, four traditional unit root tests :(KPSS、DF-GLS、ERS、NP) and Zivot and Andrew（1992）unit root with structure break to test date whether or not display stationarity. Then we use Johansen(1988, 1990, 1994) cointegration test and Gregory and Hansen(1996) cointegration test with structure break to show every variables whether have long term relationships. On the other hand, by use of Granger-causality test to display among variables whether have lead-lag relationships. Finally, this paper will use general vector autoregression model including general impulse response function and general variance decomposition to investigate all variables with Taiwan stock index open price and Taiwan election stock index open price to understand short run dynamic analysis.
For this study we can conclude below: first, we can find that from the traditional unit root tests all variables have inconsistency situation of I(1) and I(0),except for S&P 500 sock index close price. To solving this problem, while we use ZA unit root test with structure break, all variables show I(1).We can evident that all variables with stationatity. The results of Johansen cointegration test indicate that all variables with Taiwan stock index open price are considered cointegrated with one cointegrating vector.But when all variables with Taiwan election stock index open price are considered cointegrated with no cointegrating vector. Further, when using GH cointegration test with structure break to show every variables with Taiwan stock index open price or Taiwan election stock index open price have long term relationships and in the short run to display random walk situations. The results of Granger causality test significantly show that Taiwan stock index open price was led by Taiwan futures stock index close price and S&P 500 stock index close price; in the other hand about Taiwan election stock index open price was only led by Taiwan election futures stock index close price. Moreover, in the general impulse response function, we could understand that Taiwan stock index open price was impulsed by Taiwan futures stock index close price and S&P 500 stock index close price, which have more effect in Taiwan futures stock index close price than in S&P 500 stock index close price; For election, Taiwan election stock index open price was impulsed by Taiwan election futures stock index close price and NASDAQ stock index close price, which have more effect in Taiwan election futures stock index close price than in NASDAQ stock index close price, too. Finally, in the variance decomposition, we conclude that the same results of general impulse response function.
|
|---|
