摘要
本文选取五个主流的汇率基本面模型,使用2005年汇改后的人民币兑美元、欧元、日元、英镑汇率数据进行样本内拟合和样本外预测,并通过计算损失函数和SPA统计量比较五种模型的预测能力。实证结果表明:随机游走模型短期内具有更优的预测能力,但中长期内,汇率基本面模型具有更优的预测能力;总体来讲,汇率基本面模型的预测能力优于随机游走模型,人民币汇率不存在"汇率失联之谜";对不同的货币,具有最优预测能力的模型不同。
Selecting five fundamental exchange rate models, this paper fits and forecasts the RMB exchange rate with the data in-sample and out-sample after 2005. Furthermore, this paper compares the abilities of different models used to forecast the RMB exchange rate through loss function and SPA test. The results show that random walk model has better predictability in short-run, while in the medium-to-long term fundament exchange rate mod- els forecast better. In general, fundamental exchange rate models have better predictability and exchange rate of RMB is close to economic fundamentals. For different currencies, there exist different models to fit and forecast them.
出处
《数量经济技术经济研究》
CSSCI
北大核心
2016年第9期145-160,共16页
Journal of Quantitative & Technological Economics
关键词
人民币汇率
汇率基本面模型
损失函数
SPA检验
Exchange Rate of RMB
Fundamental Exchange Rate Model
Loss Func-tion
SPA test