[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: questions about dynamic panel data
I have been studying dynamic panel data lately. There
are a couple questions that I could not find answers
in the papers. I wish you could help me on these.
When conducting the Difference Sargan tests to see
whether some explanatory variable are endogenous,
predetermined, or exogenous, do the Sargan statistics
obtained under the different assumptions have to
accept the null hypothesis that the overidentifying
restrictions are valid for each model? Or whether the
instruments are valid or not in the two assumptions
does not matter for the Difference Sargan test itself?
If the Sargan test reject the null hypothesis that the
overidentifying restrictions are valid, does this mean
the instrument lags are too far or too close? Or this
depends on the particular question at hand? I think
what I am really asking here is which direction to go
to look for a valid model?
If the lag Yt-1, Yt-2, Yt-3, Yt-4 are all significant
in the model, but the average T is 5. If I include all
four lags, I throw out a lot of information of the
panel. If the estimated coefficients of Yt-1, Yt-2,
Yt-3, Yt-4 are 0.7, 0.5, 0.15, 0.05, is it safe to
just include two lags or does this will commit omitted
variable error and bias the estimators?
Thanks a million. Sorry for taking your time.
Yahoo! FareChase: Search multiple travel sites in one click.
* For searches and help try: