Testing for Regression Coefficient Stability with a Stationary AR(1) Alternative
提出一种检验回归系数是否随时间变化的方法，备择假设假设系数服从平稳AR(1)过程，并给出基于得分统计量的检验统计量，通过蒙特卡洛模拟研究其功效和大小，并以黄金和白银市场为例进行实证分析。
A bstract-We discuss the problem of testing for constant versus time varying regression coefficients. Our alternative hypothesis allows the coefficients to follow a stationary AR(1) process with unknown autoregressive parameter. Standard testing procedures are inappropriate since this parameter is identified only under the alternative. We propose a test statistic which is a function of a sequence of Score statistics, and depends only on the regressors and the OLS residuals. The distribution of the test statistic is discussed, power and size are investigated using Monte Carlo methods, and an empirical example investigating stability in the gold and silver markets is presented.