Actually, there is not big difference. The basic theory is exactlly same. When you use hausman test to compare FEM and REM in panel data, the null hypothesis is that no variables are endogeneous--x'* ε=0 , the alternative hypothesis is x`*ε≠0.
Generally, if you want to test the null hypothesis that the error terms are uncorrelated with all the regressors against the alternative that they are correlated with some of the regressors(endogenity), you can use DWH(Durbin Wu Hausman) test.
At first, run OLS and IV estimator.
Under H1, the IV estimator is consistent, but OLS estimator is not.
Under H0, Both are consistent, and OLS are more efficient.
Thus, plim(BIV-BOLS)=0 under null and nonzero under the alternative.
. reg dependentvar independentvar1 independentvar2 independentvar3 ... (OLS)
. estimates store OLS
. reg dependentvar independentvar1 independentvar2 independentvar3 ... , (IV)
(or ivreg ----------)
. estimates store IV
. hausman IV OLS
[此贴子已经被作者于2006-4-20 11:11:24编辑过]
扫码加好友，拉您进群
全部版块
我的主页
收藏
[em04]
[em09]
[em10]
