摘要
购买力平价理论是汇率预测的重要方法。应用协整理论对人民币/美元汇率的购买力平价形式进行检验可以发现,购买力平价理论一定时期内在人民币汇率上不成立。应用神经网络技术构造的非线性协整检验,结果显示其预测的效果要明显优于随机游动模型。因此可以认为在引入了非线性的机制后,基于汇率理论的结构式模型在预测汇率变化的能力上有所提高。
In this paper, with the application of co-integration technology we discuss the purchase power parity of RMB/USD exchange rate. The result we get is that PPP theory does not hold on RMB/USD exchange rate. We construct an ANN nonlinear co-integration test and test the relationship of exchange rate and CPIs of the two countries. At the end, we forecast the exchange rate using ANN models and compare the results with random walk models.
出处
《吉林大学社会科学学报》
CSSCI
北大核心
2005年第3期87-93,共7页
Jilin University Journal Social Sciences Edition
关键词
购买力平价理论
汇率预测
非线性协整
神经网络
purchase power parity
forecasting of exchange rate
nonlinear co-integration
artificial neural networks