人民币汇率的变动分析及组合预测

发布时间：2018-05-16 15:09

本文选题：汇率预测 + ARIMA模型　；参考：《湖南大学》2014年硕士论文

【摘要】：目前,汇率的分析与预测是当今经济科学领域研究的热点之一.自2005年7月21日人民币汇率形成机制改革之后，汇率浮动区间加大，这样引起的汇率风险远远大于以往.又因为美元在一篮子货币中占据着重要的地位，，正确分析与预测人民币兑美元汇率走势对我国政府正确制定货币政策、金融机构规避外汇风险有着非常重要的意义.因此本文将建立模型对人民币兑美元汇率进行走势分析和预测研究.本文研究的主要内容如下: 1.本文首先对人民币汇率波动情况进行分析.阐述了五种汇率决定理论：国际借贷学说、购买力平价学说、利率平价学说、货币主义学说和资产组合学说，并总结了影响汇率变动的各项因素如国民收入、国际收支、通货膨胀率和相对利率等等. 2.其次，本文以ARIMA模型、GARCH模型和灰色Markov模型为基础，以基于算术平均、预测误差平方和最小化和可变权数三种不同的赋权方式对人民币兑美元汇率建立非平稳时间序列的组合预测模型，对2010年7月至2013年12月的181个周平均汇率进行实证分析.根据实证研究需要，将数据分为两部分：2010年7月至2013年6月的前154个数据用来估计预测模型的参数，余下的27个数据则用来检验模型的预测效果. 3.最后本文通过采用四种误差指标度量各种预测模型的精确度，进而说明模型预测效果的优劣.结果显示基于可变权数的组合预测模型的预测精度好于其它各个预测模型.
[Abstract]:At present, the analysis and prediction of exchange rate is one of the hotspots in the field of economic science. Since the reform of RMB exchange rate formation mechanism on July 21, 2005, the floating range of RMB exchange rate has been enlarged, and the exchange rate risk caused by this increase is much greater than that in the past. Because the dollar occupies an important position in a basket of currencies, it is of great significance for our government to correctly analyze and forecast the trend of RMB exchange rate against the dollar and for financial institutions to avoid foreign exchange risk. Therefore, this paper will establish a model to analyze and forecast the trend of RMB / US dollar exchange rate. The main contents of this paper are as follows: 1. This article first carries on the analysis to the RMB exchange rate fluctuation situation. This paper expounds five kinds of exchange rate determination theories: international lending theory, purchasing power parity theory, interest rate parity theory, monetarism theory and portfolio theory, and summarizes the factors influencing exchange rate change, such as national income, international balance of payments, etc. Inflation rate and relative interest rate etc. 2. Secondly, this paper is based on ARIMA model and grey Markov model, and based on arithmetic average. This paper establishes a combination forecasting model of non-stationary time series for RMB / US dollar exchange rate in three different weighting ways of squared sum of prediction error minimization and variable weight, and empirically analyzes the average exchange rate of 181 weeks from July 2010 to December 2013. According to the need of empirical research, the data are divided into two parts: the first 154 data from July 2010 to June 2013 are used to estimate the parameters of the prediction model, and the remaining 27 data are used to test the prediction effect of the model. 3. Finally, this paper uses four kinds of error indexes to measure the accuracy of various prediction models, and then explains the advantages and disadvantages of the model prediction effect. The results show that the prediction accuracy of the combined prediction model based on variable weights is better than that of other prediction models.
【学位授予单位】：湖南大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F224;F832.6

【相似文献】