Question

Eviews6.0 面板数据Hausman检验 大神帮忙看看啊，不懂啊

Test Summary Chi-Sq. Statistic Chi-Sq. d.f. Prob.

Cross-section random 0.000000 3 1.0000
Period random 0.000000 3 1.0000
Cross-section and period random 0.000000 3 1.0000
* Cross-section test variance is invalid. Hausman statistic set to zero.
* Period test variance is invalid. Hausman statistic set to zero.
** Warning: estimated period random effects variance is zero.

首先这是什么意思啊？怎么解决或者用不用处理啊？
Cross-section random effects test comparisons:

Variable Fixed Random Var(Diff.) Prob.

LOG(LPI?) 2.391728 1.954131 -0.020151 NA
LOG(KPI?) 2.109488 2.864985 0.040001 0.0002
LOG(TE?) -2.275732 -3.112851 -0.128852 NA

Period random effects test comparisons:

Variable Fixed Random Var(Diff.) Prob.

LOG(LPI?) -0.197024 1.954131 0.109492 0.0000
LOG(KPI?) 1.156332 2.864985 1.066771 0.0981
LOG(TE?) -0.969818 -3.112851 1.357578 0.0659

Cross-section and period random effects test comparisons:

Variable Fixed Random Var(Diff.) Prob.

LOG(LPI?) -1.164758 1.954131 0.387537 0.0000
LOG(KPI?) 3.115766 2.864985 4.383856 0.9047
LOG(TE?) -2.430236 -3.112851 4.639908 0.7513

其次，这个结果是说明我应该建立个体随机模型还是个体固定效应模型啊？如果您知道，麻烦说的详细点，哪个检验结果决定建什么模型啊。为什么会出现NA啊

Answer 1

做豪斯曼之前，要先做随机效应分析，注意，option选项里选ordinary
option选项里面选择的估计方法不对，会对后面的检验产生很大影响
你这种情况应该就是估计方法设定不对导致的，可以尝试调整一下

Answer 2

hausman 检验值大于0.05 接受random假设

Answer 3

要一步一步的来做啊，你前期的单位根结果呢？
前期如果错误了，后面的结果都不用去看了