The Causality in Cointegrated Processes: Evidences on Exchange Rate and Interest Rate

• Main Authors: Jeng-Shyng Shiau, 蕭政行
• Other Authors: Ming-Chih Lee
• Format: Others
• Language: zh-TW
• Published: 2003
• Online Access: http://ndltd.ncl.edu.tw/handle/45440990407432645230

碩士 === 淡江大學 === 財務金融學系 === 91 === We always neglect cointegrated effect in Granger causality. If cointegration exist in time series and we don’t consider it, using differential series do the Granger causality straightforwardly, or even consider it but using ECM does, it maybe have some problems. This studies using another method to do the Granger causality developed by Toda and Yamamoto (1995). According to Toda and Yamamoto, needn’t under unit root and cointegration test, Wald test statistics still asymptotic distribution. According to Hatemi-J and Irandoust (2000), exchange rate and interest rate will be feedback causality if a country’s international capital is mobile. In our study, the situation really happens in Toda and Yamamoto’s causality test. But if we neglect cointegrated effect in Granger causality, it is just exchange rate causes interest rate. So, considering and non-considering cointegration may lead to different conclusions. This study use rolling method to forecast 2002:12 exchange rate wanting to arbitrage. Although considering and non-considering cointegration ways forecasting exchange rate do not succeed in arbitraging, considering cointegration way has smaller deviation standard error than non-considering’s. So considering cointegration method is a better way to arbitrage than non-considering.