本质上你没清楚hausman检验的原理。

你只要交换两种模型的估计次序，就可以将负变为正，绝对值不变

楼上的大虾,能解释一下如何交换次序吗?感激不尽!

谢谢你 的提醒

楼上可能把HAUSMAN TEST 为负 想的简单了一些。You can not exchange the order of hausman test. By original order, the variance-covariance matrix muse be PD. That is, by theory hausman test statistic is positive. But this is large sample property,in empircal study, no reason believes the variance-covariance matrix is PD. If you change the order, it is not the HAUSMAN test and you can not get any useful results .

i post a paper about negative hausman test statistic. please take a look.

[此贴子已经被作者于2007-4-19 4:14:49编辑过]