Likelihood ratio test (LR) for identify the individual firm effect

[priya sawaliya]

Hello,

I want to test the joint significance of the individual effects in my fixed effect model and for this I run the following command :

xtreg y x1 x2 x3 x4, fe
estimates store m1

xtreg y x1 x2 x3 x4 i*, fe
estimates store m2

lrtest m1 m2

I get the error: df(unrestricted) = df(restricted)

How shloud I do this. Please help

Thank you
Priya

[Joseph Coveney]

How shloud I do this. Please help

You'll get better help by following the FAQ.

.ÿ
.ÿversionÿ15.1

.ÿ
.ÿclearÿ*

.ÿ
.ÿquietlyÿsysuseÿauto

.ÿ
.ÿquietlyÿxtsetÿrep78

.ÿ
.ÿquietlyÿxtregÿgear_ratioÿc.lengthÿi.foreign,ÿfe

.ÿestimatesÿstoreÿFull

.ÿ
.ÿquietlyÿxtregÿgear_ratioÿc.length,ÿfe

.ÿlrtestÿFull

Likelihood-ratioÿtestÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿLRÿchi2(1)ÿÿ=ÿÿÿÿÿ18.59
(Assumption:ÿ.ÿnestedÿinÿFull)ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿProbÿ>ÿchi2ÿ=ÿÿÿÿ0.0000

.ÿ
.ÿexit

endÿofÿdo-file


.

I'm guessing that some collinear predictors got booted, and the degrees-of-freedom ended up the same, but you're hiding everything, so it's hard to say for sure.

Another recommendation for Stata 16: get rid of xtset if only to force people to show at least some of what they are doing when they post on Statalist.

[Joseph Coveney]

After reading the title, I'm now guessing that your i* is one or more variables that identify the same entities as whatever it was you used for xtset. So, you could either

xtset firm_id
xtreg y x1 x2 x3 x4, fe

and glance down below the regression table to the F statistic shown there.

Or, if you've got some kind of problem with that, then you could do something like the following (taken after the example above).

.ÿ
.ÿversionÿ15.1

.ÿ
.ÿclearÿ*

.ÿ
.ÿquietlyÿsysuseÿauto

.ÿ
.ÿsysuseÿauto
(1978ÿAutomobileÿData)

.ÿquietlyÿregressÿgear_ratioÿc.lengthÿi.rep78

.ÿestimatesÿstoreÿFull

.ÿ
.ÿquietlyÿregressÿgear_ratioÿc.lengthÿifÿe(sample)

.ÿlrtestÿFull

Likelihood-ratioÿtestÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿLRÿchi2(4)ÿÿ=ÿÿÿÿÿÿ6.24
(Assumption:ÿ.ÿnestedÿinÿFull)ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿProbÿ>ÿchi2ÿ=ÿÿÿÿ0.1817

.ÿ
.ÿexit

endÿofÿdo-file


.

[priya sawaliya]

Thank you Joseph sir for your reply.
There are 63 companies in my panel data and I have created dummy for each company. So this i* is representing dummy for companies.

[Joseph Coveney]

With

xtset firm_id
tabulate firm_id, generate(i)
y x1 x2 x3 x4 i*, fe

will drop all of your indicator variables and so the degrees of freedom will be the same as without, that is, will have the same degrees of freedom as

xtreg y x1 x2 x3 x4, fe

That's why you got the error message.

As I mentioned, the command automatically tests the hypothesis that you wish to, although it is done in the form of an F statistic.