2SLS , Hausman test , Sargan test

是 否 +2 论坛币

k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

Dear all :

It seems we can not just click button to get Hausman test and Sargan test in EViews' window , so I have some ideas

about that by 2SLS method .

model :  yi = xi*(beta) + ui   , i = 1,....,n

Hausman test:

Ho : xj is exogeneous , Ha : not Ho , j = 1 , ... , k , where k<n

step 1 : run y on x (all x) --> save ui(hat) &

            run xj on all z --> save vj(hat) , where z is instrument variables and vi is from : xj = z*r + vj

step 2 : run ui(hat) on  x (all x) and vj --> save R^2

step 3 : compute LM=n*R^2 ~ X^2(k) [Chi-squared distribution with degree of freedom k]

Sargan test :

Ho: z is exogeneous , Ha: not Ho

step 1 : run y on x (all x) by instrument variable method --> save residuals ui( IV hat )

step 2: run ui( IV hat ) on z --> save R^2

step3 : compute LM =n*R^2 ~X^2(m-k) [Chi-squared distribution with degree of freedom m-k]

            where m is the number of instrument variables

Hope this will help !

[此贴子已经被作者于2009-5-30 23:52:18编辑过]

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

分享0 收藏1 回帖

关键词：hausman ausman Sargan test 2SLS test hausman Sargan

谢谢，虽然没看怎么明白～

楼主解释的很清楚！
可以详细说一下compute LM =n*R^2 ~X^2(m-k)怎么操作吗？要输入command还是哪里可以直接click？

Thanks a lot,
althought Idon't understand.

1# dlovea 谢谢。貌似Eviews7提供Sargan:
J-statistic is simply the Sargan statistic

Note that these residuals are based on X, not X hat.