摘要
带滞后因变量的小样本最小二乘法参数估计会产生低估偏误,中值无偏估计是修正偏误的一个很好的方法。基于AR(1)的中值无偏估计存在确定解,本文将应用Boostrap仿真实验对计算其数值解作出尝试。本文根据1994年1月至2004年5月期间人民币实际汇率月度数据,用中值无偏估计的方法计算了人民币实际汇率单位扰动半生命期(HLS),根据仿真计算的结果为11·5个月。
The estimation of OLS with lagged independent variables has a downward bias. Median - unbiased estimation is a good method to solve such problem, which has exactly solution in the AR (1) models This paper will try to give a numerical solution based on the Bootstrap simulation. Also this paper will calculate the HLS of Chinese RMB's real exchange rate with a sample duration of 1994. 01 to 2004. 05 by using the method of Median - unbiased estimation, and gets the result of 11.5 months based on the Bootstrap simulation.
出处
《数量经济技术经济研究》
CSSCI
北大核心
2005年第11期78-85,共8页
Journal of Quantitative & Technological Economics