【作者】 杨益波；

【导师】 谢赤；

【作者基本信息】 湖南大学 ， 金融学， 2003， 硕士

【摘要】 汇率是金融市场的重要变量，对一国国民经济的健康运行有着很大的影响。自布雷顿森林体系解体后，浮动汇率制度成了世界上的主要汇率制度。汇率波动的频率与幅度越来越大，汇率风险成为金融市场上最重要的风险之一。全世界对汇率的变动给予了高度的关注，各国政府开始采取各种措施以使汇率的变动有利于本国的经济发展，各公司也想出许多方法来预防和减少因汇率波动带来的外汇风险。因此，对汇率行为和波动性的描述及在此基础上进行的汇率预测就成为了金融领域研究的热点与难点。随着计量经济学及计算机技术的发展和运用，新的方法和工具不断被运用到有关汇率的研究中来。本文采用上个世纪80年代以后开始发展起来的协整分析技术，对汇率的行为进行描述和预测。 本文首先回顾了汇率制度演变的过程及在此过程中汇率行为的表现方式的变化，特别是详细总结了当今汇率行为的特征。在回顾了经典的汇率理论及其相应的模型特别是详细讨论了货币主义汇率理论后，本文总结了传统的汇率行为描述和预测方法的不足并探讨了寻找新的方法的方向。然后，本文对协整分析技术的原理和具体方法进行了系统的阐述，提出了基于协整分析技术的汇率行为描述和预测的具体步骤。为了提高模型的准确性，本文引进人工神经网络技术建立了非线性误差校正模型，减少预测误差。实证结果说明协整分析技术是对当今汇率行为进行描述和预测的有效方法，人工神经网络能够很好的把握汇率波动中的非线性因素。更多还原

【Abstract】 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.更多还原