汇率、金价、油价、利率、物价之因果关系检定与预测——向量误差修正模型之应用

摘要

This research is examining and forecasting the casuality between exchange rate, gold price, oil price, in- terest rate, and CPI. ADF unit root test shows that after first order difference, all variables exhibit stationary state. Coin- tegrating test shows all variables have the Cointegration relationship. We find that the preceding rise of gold price will induce the drop of exchange rate and will trigger the price can usually lead to the rise of gold price, while the next couple of weeks. A rise of gold price will be falling of gold price in the near future. The preceding rises of oil current rise of interest rate will cause a decline of interest rate in followed by a decline of CPI. A rise of CPI can be expected after the climbing of interest rate. The Granger causality relationship Test shows that the gold price can Granger influence interest rate and oil price. The interest rate can Granger affect exchange rate, but exchange rate and CPI can＇t Granger affect other variables. Finally, we forecast exchanger rate and CPI by VECM, and the MAE, MSE, and MAPE all obey the criterion. The results of forecast are nn~itiv~ nn ~l~ti~t;~,~%该研究是针对汇率、金价、油价、利率、物价之因果关系检定来分析，经ADF单根检定显示：一阶差分后所有变量数据皆呈现定态，再由共整合检定得知，汇率、金价、油价、利率、物价存在共整合关系，透过VECM发现：前期金价提高将导致本期汇率下跌；前期金价提高将使本期金价下跌；前期油价提高将促使本期金价上涨；前期利率提高会使本期利率下跌；前期金价上升将使本期物价下跌；前期利率提高将造成本期物价上涨。至于Granger因果关系检定显示：金价会Granger影响利率和油价，利率会Granger影响汇率，但汇率及物价不会Granger影响其它变量。最后将汇率和物价分别与以VECM来预测，预测模型之MAE，MSE，及MAPE准则皆符合，显示预测结果良好。