[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: RE: xtivreg2, error 198
thanks a lot for your answer. Yes, I think -nonest- and -dfadj- would
suit my needs; unfortunately, I cannot realy test them, because my
Stata dates from the time you wrote the -xtivreg2-, i.e. before -xtreg-
had the cluster option. Of course, it would be fantastic if you could
put -nonest- and -dfadj- into -xtivreg2- someday.
As a first approximation, I demeaned all my variables and estimated the
model with -ivreg2- and the cluster option; I am aware that the SEs are
too small and a dof adjustment is needed. My question is, however,
whether this approach yields the same result as -xtivreg2- with the
-nonest- option would.
The problem is interesting but tricky. The usual classical and robust
var-cov matrix for the fixed effects estimator need a degrees of
freedom adjustment for the fixed effects (see any textbook treatment).
The cluster-robust var-cov matrix doesn't need this adjustment, so
long as the panels (corresp. to the fixed effects) don't span more
than one cluster.
The problem arises when the panels cut across clusters. Some dof
adjustment is needed, but what should it be? I don't know, and I
haven't seen any paper that describes exactly what it should be. When
writing -xtivreg2-, I opted simply for the program to exit with error.
Official -xtivreg- now has the -nonest- and -dfadj- options; these
didn't exist when I wrote -xtivreg2-. You'll see that they allow you
to go between the extremes of no dof adjustment (which will give you
SEs that are too small) and a full dof adjustment (which will give you
"conservative" SEs that are on the large side). Have a look at them
and see if they would suit your needs. I could perhaps put them into
Prof. Mark Schaffer
Department of Economics
School of Management & Languages
Heriot-Watt University, Edinburgh EH14 4AS
tel +44-131-451-3494 / fax +44-131-451-3296
* For searches and help try: