摘要
基于Ｋｔｕｇｍ ａｎ 的汇率目标区域理论和中国现行的外汇管理体制及汇率制度的基本特点，应用人工神经网络技术对１９９４ 年１ 月－ １９９７ 年３ 月的月度数据，建立人民币汇率的非线性模型：目标区域模型；并运用所建立的模型对１９９７ 年４ 月－ ９ 月的人民币汇率走势进行预测．结果表明该模型具有令人满意的拟合与预测能力．基于目标区域模型在中国的适用性，应用该模型的平滑过渡条件，求出中国外汇储备的最佳上限（１５７７．５ 亿美元）．在确定出人民币汇率的目标区域的基础上，提出应适度贬值人民币的政策建议，并阐述实行这一举措的可行性和必要性．
Based on the Krugman target zone model and foreign exchange control and manage floating system of China, this paper employs an artificial neural networks to construct a non\|linear model of RMB exchange rate, i.e., the exchange rate target zone model. The monthly data for the period form January 1994 to March 1997 are used for modeling and the following six months in 1997 for forecast. The results show that the model fits the actual well and has good forecast power. Based on RMB exchange rate target zone model, the optimal upper limit of foreign reserves size of China is resolved as well by making use of the model’s “smooth pasting” condition.
出处
《系统工程理论与实践》
EI
CSCD
北大核心
1999年第10期55-62,共8页
Systems Engineering-Theory & Practice
基金
国家自然科学基金
关键词
人民币汇率
外汇储备
目标区域模型
汇率制度
RMB exchange rate
foreign exchange reserves
exchange rate target zone model
artificial neural networks
