Font Size: a A A

Forecasting RMB Exchange Rate Based On The Nonlinear Combination Model Of ARFIMA-SVM-BPNN

Posted on:2016-12-05Degree:MasterType:Thesis
Country:ChinaCandidate:Z MaoFull Text:PDF
GTID:2349330473965943Subject:Management Science and Engineering
Abstract/Summary:PDF Full Text Request
On July 21, 2005, RMB exchange rate began to implement the managed floating exchange rate system based on market supply and demand, reference to a basket of currencies. As the mutual penetration and influence of international economic and financial activities, the degree of interdependence between financial markets of different countries also gradually enhanced. As the nuclear variable, exchange rate adjusts and affects the micro and macro economy from country to country. Since China’s accession to the WTO in 2001, the position of China’s economy in the international community has become more important. As the bridge associating the national and international economy, RMB exchange rate becomes the focus of attention. Forecasting the RMB exchange rate with the combination of theory and demonstration not only has the important theoretical significance, but provides a new view of making corresponding policy to the government management department.This paper first summarizes the methods of forecasting the RMB exchange rate time series, then introduces and analyzes exchange rate determination theory and the testing method of nonlinear characteristics. After that, the paper builds the ARFIMA model, the SVM model and BPNN model by analyzing the long memory and nonlinear characteristics of RMB exchange rate, and constructs a nonlinear combination model combining different single models. Finally RMB/USD and RMB/EUR are chosen as samples; and the paper uses BDS and DFA model to analyze the nonlinear characteristic and long memory characteristic of the RMB exchange rate, and then forecast s the RMB exchange rate by the nonlinear combination model. Comparing the effectiveness of different models, the paper puts forward the corresponding policy suggestions.The research results show that in terms of nonlinear test and long memory test, the two RMB exchange rate yield sequences have nonlinear characteristics, and the RMB/USD yield has the strong long memory, the RMB/EUR yield has the weak long memory. In terms of nonlinear combination model construction and application, the RMB exchange rate predictive effect of nonlinear combination based on ARFIMA, SVM, BPNN models is better than that of single models.
Keywords/Search Tags:Exchange rate yield, ARFIMA model, SVM model, BPNN model, Nonlinear combination model
PDF Full Text Request
Related items