發(fā)布時間：2019-07-08 10:38

【摘要】：本文針對匯率問題的復(fù)雜性,根據(jù)匯率與宏觀經(jīng)濟變量的非線性關(guān)系和匯率數(shù)據(jù)自身的非線性特征,以神經(jīng)網(wǎng)絡(luò)技術(shù)為主要工具,研究了人民幣匯率基于購買力平價理論的預(yù)測和時間序列預(yù)測問題,并建立了一個基于神經(jīng)網(wǎng)絡(luò)的人民幣匯率風險管理的VaR 模型。 本文應(yīng)用神經(jīng)網(wǎng)絡(luò)考察匯率與物價指數(shù)之間的非線性關(guān)系。在應(yīng)用協(xié)整技術(shù)檢驗發(fā)現(xiàn)購買力平價理論對人民幣匯率不成立的情況下,論文建立了一種基于神經(jīng)網(wǎng)絡(luò)的非線性協(xié)整檢驗方法,對匯率與物價指數(shù)進行了非線性協(xié)整檢驗。并在此基礎(chǔ)上建立了預(yù)測模型。 論文在分析人民幣匯率收益序列長記憶性的基礎(chǔ)上,建立了神經(jīng)網(wǎng)絡(luò)預(yù)測模型。引入方向準確率作為預(yù)測模型的一個評價指標,對模型的預(yù)測結(jié)果進行了統(tǒng)計檢驗。并考察了計量經(jīng)濟模型的選擇標準中的信息準則法,在建立神經(jīng)網(wǎng)絡(luò)預(yù)測模型過程中的作用。 論文根據(jù)神經(jīng)網(wǎng)絡(luò)技術(shù)中的混合密度網(wǎng)絡(luò)能夠預(yù)測數(shù)據(jù)之間的條件概率密度函數(shù)這一特點,對外匯資產(chǎn)組合收益的條件密度函數(shù)進行了預(yù)測,并在預(yù)測的基礎(chǔ)上建立了一個VaR 模型。 論文比較系統(tǒng)地分析了人民幣匯率的特征,考察了應(yīng)用神經(jīng)網(wǎng)絡(luò)技術(shù)進行匯率預(yù)測的建模過程和模型選擇技術(shù),討論了對人民幣匯率進行有效預(yù)測和風險管理的可行性。這些工作,將有助于拓寬對人民幣匯率問題的研究思路,為解決人民幣匯率預(yù)測和風險管理問題提供了一些可供選擇的方法。
文內(nèi)圖片：
圖片說明：-1資本市場的均衡圖中ME表示貨幣市場的均衡曲線，，它是向上傾斜的
[Abstract]:In view of the complexity of exchange rate problem, according to the nonlinear relationship between exchange rate and macroeconomic variables and the nonlinear characteristics of exchange rate data itself, this paper studies the prediction and time series prediction of RMB exchange rate based on purchasing power parity theory with neural network technology as the main tool, and establishes a VaR model of RMB exchange rate risk management based on neural network. In this paper, neural network is used to investigate the nonlinear relationship between exchange rate and price index. When it is found that purchasing power parity theory is not valid for RMB exchange rate by using cointegration technique, a nonlinear cointegration test method based on neural network is established in this paper, and the nonlinear cointegration test of exchange rate and price index is carried out. On this basis, the prediction model is established. Based on the analysis of the long memory of RMB exchange rate return series, a neural network prediction model is established in this paper. The direction accuracy is introduced as an evaluation index of the prediction model, and the prediction results of the model are statistically tested. The role of the information criterion method in the selection criteria of econometric models in the process of establishing neural network prediction model is also investigated. According to the characteristic that the mixed density network in neural network technology can predict the conditional probability density function between data, this paper forecasts the conditional density function of foreign exchange asset portfolio income, and establishes a VaR model on the basis of prediction. This paper systematically analyzes the characteristics of RMB exchange rate, investigates the modeling process and model selection technology of applying neural network technology to exchange rate prediction, and discusses the feasibility of effective prediction and risk management of RMB exchange rate. These work will help to broaden the research ideas of RMB exchange rate and provide some alternative methods to solve the problems of RMB exchange rate forecasting and risk management.
【學(xué)位授予單位】：吉林大學(xué)
【學(xué)位級別】：博士
【學(xué)位授予年份】：2005
【分類號】：F832.6

【引證文獻】

相關(guān)期刊論文 前1條

1 張晨曦;楊一文;;基于混合密度網(wǎng)絡(luò)測度股市流動性“周內(nèi)效應(yīng)”[J];中國證券期貨;2010年09期

相關(guān)碩士學(xué)位論文 前9條

1 鄭東;我國商業(yè)銀行匯率風險形成機制及其管理研究[D];廣東外語外貿(mào)大學(xué);2006年

2 張靜;開放條件下我國商業(yè)銀行匯率風險管理問題研究[D];湖南大學(xué);2006年

3 鄧雄;我國商業(yè)銀行匯率風險管理研究[D];西南財經(jīng)大學(xué);2007年

4 王映喬;基于神經(jīng)網(wǎng)絡(luò)的匯率預(yù)測及系統(tǒng)設(shè)計[D];西南財經(jīng)大學(xué);2007年

5 周文;基于密度預(yù)測的人民幣匯率組合預(yù)測研究[D];中南大學(xué);2006年

6 姜瑩;匯率體制轉(zhuǎn)軌時期我國商業(yè)銀行外匯風險管理研究[D];南京師范大學(xué);2008年

7 馬廣元;我國商業(yè)銀行匯率風險管理研究[D];西南財經(jīng)大學(xué);2009年

8 盧寧;企業(yè)外匯風險防范決策支持系統(tǒng)[D];河南科技大學(xué);2012年

9 侯英超;基于VaR的中國股指期貨風險實證研究[D];北方工業(yè)大學(xué);2013年

本文編號：2511530