本研究之研究對象為臺灣上市金融類股價指數,藉由時間序列方法中的ARIMA 及 GARCH 模式,對 2005 年 9 月至 2015 年 7 月間之月資料進行分析、預測,並比較兩模型之預測績效,再以 VAR 模型分析總體經濟變數及國際股市對臺灣上市金融類股價指數之影響,有助於確定臺灣上市金融類股價指數與各變數之間的連動性,期能更準確的預測臺灣上市金融類股價指數未來的走向。 本研究經實證結果顯示,在預測模式中,以 ARIMA(1,1,1) 為最佳模式;在關聯性分析中,由 Granger 因果關係檢定結果顯示,臺灣上市金融類股價指數與利率、領先指標、同時指標指數及日本東京日經225指數間存在著互為因果之回饋關係;由衝擊反應分析可知,當總體經濟變數及國際股市發生變動時,臺灣上市金融類股價指數對利率、領先指標指數及同時指標指數之反應最為明顯;由預測誤差變異數分解可知,臺灣上市金融類股價指數主要被利率、領先指標指數、同時指標指數及美國標準普爾股價指數等變數所解釋。
The object of the study is the stock prices index of financial in Taiwan.We collect the data between September 2005 to July 2015, then analyze and compare the predictive performance with ARIMA of time series and GARCH models. After that, we use VAR model to analyze the impact between macroeconomic variables, international stock market and the stock prices index of financial in Taiwan. It would be helpful to ensure the linkage between the stock prices index of financial in Taiwan and each variable, then we hope to predict the future direction of the stock prices index of financial in Taiwan more accurately. The empirical results show that the ARIMA(1,1,1) is the best model in the forecasting models. In correlation analysis, the Granger causality test results show that the financial stock index and interest rate, leading indicators, coincident indicators index and the Nikkei 225 index in Tokyo, Japan there are both cause and effect relationship of feedback. In impulse response, when macroeconomic variables and international stock market are changed, the response between the stock prices index of financial in Taiwan and interest rate, leading indicators, coincident indicators index is the most obvious. The analysis of variance decomposition shows that the stock prices index of financial in Taiwan is mainly explained by interest rate, leading indicators, coincident indicators index, the S&P 500 index in USA and other variables.