HOW to obtain the log-likelihood values in a pooled OLS and FE estimations (panel data) ????

[Kodi Hannon]

Hello everybody!

My issue is that in some papers using panel data, I noticed that in the estimate results inherent a pooled OLS regression, they report the value of the log-likelihood.
I was wondering how this is possible in Stata, since OLS and ML are two separate estimators (leaving apart the normality assumption of the residuals to make them coincide).

For instance, to be more clear, see this paper (pag. 16, Table 1):
http://ageconsearch.umn.edu/bitstrea...ing%20Zhao.pdf
How were the log likelihood values obtained in the panel estimates of Table 1 through the implementation of POLS, and FE estimators?..which is the command to obtain them?


ps.: a second (smaller) clarification: I suppose that sigma squared in Table 1 refers to the Root MSE, right?
Thank you very much!!

Kodi

[Carlo Lazzaro]

Kodi:
you may want to take a look at -Example 5: Random-effects model fit using ML- under -xtreg- entry in Stata .pdf manual.

[Kodi Hannon]

Hi Carlo,

thank you for your reply

Yes, I know about the xtreg ..., mle command, however, that command solely refers to the random effect..what about the fixed effect?

And, above all, what about the pooled OLS regression? ("reg", not "xtreg"). I didn't find anything on this..

[Carlo Lazzaro]

Kodi:
I'm not aware of MLE for -xtreg, fe-.
As far as you second question is concerned, you may want to take a look at the following link http://stats.stackexchange.com/quest...ing-mle-vs-ols

[Joseph Coveney]

I'm not sure what "pooled OLS regression" means, but if you're going

regress response . . . i.id

, then you can get the corresponding log-likelihood by going

glm response . . . i.id

But if you want the log-likelihood with a fixed-effects panel model fitted with xtreg . . ., fe (which is uses ordinary least-squares estimation as far as I am aware), then use estat ic afterwards.

Edited to add: you can also use estat ic after regress, too.

.ÿversionÿ14.0

.ÿ
.ÿclearÿ*

.ÿsetÿmoreÿoff

.ÿ
.ÿsysuseÿbplong
(fictionalÿblood-pressureÿdata)

.ÿ
.ÿxtregÿbpÿi.when,ÿi(patient)ÿfe

Fixed-effectsÿ(within)ÿregressionÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿNumberÿofÿobsÿÿÿÿÿ=ÿÿÿÿÿÿÿÿ240
Groupÿvariable:ÿpatientÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿNumberÿofÿgroupsÿÿ=ÿÿÿÿÿÿÿÿ120

R-sq:ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿObsÿperÿgroup:
ÿÿÿÿÿwithinÿÿ=ÿ0.0856ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿminÿ=ÿÿÿÿÿÿÿÿÿÿ2
ÿÿÿÿÿbetweenÿ=ÿÿÿÿÿÿ.ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿavgÿ=ÿÿÿÿÿÿÿÿ2.0
ÿÿÿÿÿoverallÿ=ÿ0.0380ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿmaxÿ=ÿÿÿÿÿÿÿÿÿÿ2

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿF(1,119)ÿÿÿÿÿÿÿÿÿÿ=ÿÿÿÿÿÿ11.14
corr(u_i,ÿXb)ÿÿ=ÿ0.0000ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿProbÿ>ÿFÿÿÿÿÿÿÿÿÿÿ=ÿÿÿÿÿ0.0011

------------------------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿbpÿ|ÿÿÿÿÿÿCoef.ÿÿÿStd.ÿErr.ÿÿÿÿÿÿtÿÿÿÿP>|t|ÿÿÿÿÿ[95%ÿConf.ÿInterval]
-------------+----------------------------------------------------------------
ÿÿÿÿÿÿÿÿwhenÿ|
ÿÿÿÿÿÿAfterÿÿ|ÿÿ-5.091667ÿÿÿ1.525736ÿÿÿÿ-3.34ÿÿÿ0.001ÿÿÿÿ-8.112776ÿÿÿ-2.070557
ÿÿÿÿÿÿÿ_consÿ|ÿÿÿÿÿ156.45ÿÿÿ1.078858ÿÿÿ145.01ÿÿÿ0.000ÿÿÿÿÿ154.3138ÿÿÿÿ158.5862
-------------+----------------------------------------------------------------
ÿÿÿÿÿsigma_uÿ|ÿÿ9.7739785
ÿÿÿÿÿsigma_eÿ|ÿÿ11.818301
ÿÿÿÿÿÿÿÿÿrhoÿ|ÿÿ.40616276ÿÿÿ(fractionÿofÿvarianceÿdueÿtoÿu_i)
------------------------------------------------------------------------------
Fÿtestÿthatÿallÿu_i=0:ÿF(119,ÿ119)ÿ=ÿ1.37ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿProbÿ>ÿFÿ=ÿ0.0444

.ÿestatÿic

Akaike'sÿinformationÿcriterionÿandÿBayesianÿinformationÿcriterion

-----------------------------------------------------------------------------
ÿÿÿÿÿÿÿModelÿ|ÿÿÿÿÿÿÿÿObsÿÿll(null)ÿÿll(model)ÿÿÿÿÿÿdfÿÿÿÿÿÿÿÿÿAICÿÿÿÿÿÿÿÿBIC
-------------+---------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿ.ÿ|ÿÿÿÿÿÿÿÿ240ÿ-859.8148ÿÿ-849.0792ÿÿÿÿÿÿÿ2ÿÿÿÿ1702.158ÿÿÿÿ1709.12
-----------------------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿNote:ÿN=ObsÿusedÿinÿcalculatingÿBIC;ÿseeÿ[R]ÿBICÿnote.

.ÿ
.ÿexit

endÿofÿdo-file


.

[Kodi Hannon]

Thank you all.

And thank you Joseph, you are a king!