<P>We all know that if an AR(p) process is stationary(covariance stationary, or asymptotically stable), the ML estimator must be consistent! What if the process is NOT stationary? We assume that the weak exogeneity still holds. Is the ML estimator still consistent? Why? State your PROOF, please!</P>
<P></P>
<P>中文:</P>
<P>如果p阶自回归时间序列是平稳的(协方差平稳,或者渐进稳定),那么其极大似然估计量就是一致的。但是如果该时间序列过程不是平稳(协方差平稳,或者渐进稳定)的,会怎么样呢?我们假定所有自变量相对所有待评估参数都是弱外生的。极大似然估计量依旧一致么?为什么?试讨论并给出相应证明。</P>
[此贴子已经被作者于2006-10-20 1:56:48编辑过]