[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: gmm with a huge panel

From   "jenny.montaldo" <>
To   "statalist" <>
Subject   st: gmm with a huge panel
Date   Mon, 4 Feb 2008 18:27:15 +0100

Dear all, 

I have a panel of 400000 observations, for like 80000 individuals over time. I performed fixed effects and ols estimations. Then I tried to perform some gmm estimates. It is quite difficult since my pc allows only to perform them either with the collapse command or with just one lag (3 3) for instance. 
The results display often quite bigger coefficient (significant) than normal ols and fixed effects estimates. Moreover the sargan test never passes and the hansen test says it is robust, but can be weakend by many instruments, when in reality it does not pass and the instruments are much lower than the number of groups. (since I can not try with more instruments). 
Can this problem be due to the relatively high number of observations and few number of instruments?
can anyone help me or suggest me something?


*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index