Describing And Forecasting Exchange Rate Behavior Based On Cointegration Analysis

Exchange rate has an important effect on macro-economy. It is an important variable of financial market. After the breakdown of the Bretton Woods System, floating exchange rate system became the main exchange rate system. Exchange rate fluctuates frequently, rising and falling uncertainly. Exchange rate risk became one of the most important financial risks. It has been given much attention and all countries and many companies have taken many measures in order to make the best of exchange rate fluctuation. So, the analysis between exchange rate and other macroeconomic variables and behavior description and forecasting of exchange rate based on it are attractive in financial field. With the development of computer technology and econometrics, new ways and tools have been used to study about exchange rate fluctuation. This paper uses the cointegration analysis technology which has been widely used from 80's of last century to describe exchange rate behavior and forecast it.The paper firstly reviews the change process of exchange rate system and analyzes the character of exchange rate behavior. After reviewing traditional exchange rate theories and their models especially monetarism exchange rate theory, the paper analyzes the shortnesses of the traditional exchange rate behavior describing and forecasting ways and discusses the direction to look for the new method. Then this paper expatiates the theory and concretely technology about cointegration. Based on this, the paper puts forward the correctly steps to describe exchange rate and forecast it. In order to improve the model's veracity, the paper uses artificial neural network to establish nonlinear error correction model. Demonstration results has proved that cointegration analysis is a good way to describe and forecast exchange rate and artificial neural network can catch the nonlinear factors of exchange rate fluctuation.