Bootstrap confidence intervals in a switching regressions model

被引：5
|
作者：
Douglas, S
机构：
[1] Department of Economics, West Virginia University, Morgantown, WV 26506-6025
来源：
ECONOMICS LETTERS | 1996年 / 53卷 / 01期
关键词：
resampling; change-point; percentile; parameter shift;
D O I：
10.1016/S0165-1765(96)00880-4
中图分类号：
F [经济];
学科分类号：
02 ;
摘要：
Traditional switching regression methods produce slope and intercept estimates conditional on the change point estimate, with confidence intervals that overstate their precision. This paper describes the problem and a bootstrap alternative. Extensive sampling experiments confirm that the traditional methods overstate precision, and that bootstrap confidence intervals are far more accurate.
引用
收藏
页码：7 / 15
页数：9
相关论文
共 50 条
12345