Hello again.
This is really strange: I actually get the right numbers all the time. I tried
now and got
> data("EmplUK", package="plm")
> zz <- pggls(log(emp)~log(wage)+log(capital),data=EmplUK, model="random")
Warning message:
'random' argument to pggls() has been renamed as 'pooling'
> summary(zz)
Random effects model
Call:
pggls(formula = log(emp) ~ log(wage) + log(capital), data = EmplUK,
model = "random")
Unbalanced Panel: n=140, T=7-9, N=1031
Residuals
Min. 1st Qu. Median Mean 3rd Qu. Max.
-1.8440000 -0.3908000 0.0388700 0.0005525 0.4153000 1.4920000
Coefficients
Estimate Std. Error z-value Pr(>|z|)
(Intercept) 1.751448 0.180100 9.7249 < 2e-16 ***
log(wage) -0.132986 0.054292 -2.4495 0.01431 *
log(capital) 0.629621 0.018483 34.0648 < 2e-16 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Total Sum of Squares: 1853.6
Residual Sum of Squares: 386
Multiple R-squared: 0.79176
exactly as expected.
Which is your system? (please post the output of Sys.info) .... although I'd be
surprised if this were system dependent!
You might also update the package to the last version if it isn't yet, although
I can't remember bugs in this function since a very long time.
Thx for feedback
Best,
G.
________________________________
Da: Ruben de Bliek [mailto:rubendebl...@gmail.com]
Inviato: venerdì 27 aprile 2012 16.28
A: Millo Giovanni
Cc: r-help@r-project.org
Oggetto: Re: [R] PLM package PGGLS strange behavior
Thank you Millo. I was a little confused by the random versus pooling
nomenclature used in PLM, thank you for clearing that up. I still have the
issue of not receiving the proper coefficient estimates for the example in the
paper though. My output is posted below; the estimates are substantially
different from the ones posted on page 20. My R version is 2.14.2.
library(plm)
> data("EmplUK", package="plm")
> zz <- pggls(log(emp)~log(wage)+log(capital),data=EmplUK, model="random")
Warning message:
'random' argument to pggls() has been renamed as 'pooling'
> summary(zz)
Random effects model
Call:
pggls(formula = log(emp) ~ log(wage) + log(capital), data = EmplUK,
model = "random")
Unbalanced Panel: n=140, T=7-9, N=1031
Residuals
Min. 1st Qu. Median Mean 3rd Qu. Max.
-1.80700 -0.36550 0.06181 0.03230 0.44280 1.58700
Coefficients
Estimate Std. Error z-value Pr(>|z|)
(Intercept) 2.023480 0.158468 12.7690 < 2.2e-16 ***
log(wage) -0.232329 0.048001 -4.8401 1.298e-06 ***
log(capital) 0.610484 0.017434 35.0174 < 2.2e-16 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Total Sum of Squares: 1853.6
Residual Sum of Squares: 402.55
Multiple R-squared: 0.78283
2012/4/27 Millo Giovanni <giovanni_mi...@generali.com>
Hello. No "strange behaviour" here, just a warning.
There is a difference between an "error" and a "warning", and between an
argument and a model. In this specific case, the warning is just there
to remind you that, as stated, 'the "random" **argument** has been
renamed to "pooling" ' (emphasis mine).
Both still work, but the former is deprecated. The estimator you get is
the same (a GGLS, or "Parks estimator"), everything works the way it
should and you can trust the numbers that come out, provided the
specification is clear to you and it is what you wanted (which for
instance is not entirely clear from your email).
The model underlying the General Feasible GLS estimator (GGLS) does not
really have "random effects"; therefore, after having initially named
the model without FEs "random" by contrast to the Fixed Effects GLS a la
Kiefer (1980), we later considered the denomination as inappropriate and
changed it to "pooling" which in effect it is: a pooled model with no
proper individual effects but a general error covariance structure. You
can get a better understanding of the specification if you read the
cited reference carefully (or even better, Wooldridge as referenced
therein).
Lastly, it is not clear what you mean by "the right estimates": the
numbers I get by using either 'model="pooling"' or ' model="random"' on
the given example are exactly those in the JSS paper. If on your system
you get anything else, I'll be grateful for a reproducible report, as
asked for in the posting guide.
PS if by chance you are just mistaking "General Feasible GLS" with
"Random Effects by GLS", then you should use 'plm(yourformula, yourdata,
model="random")' instead, and you'll get the standard RE model.
Best wishes,
Giovanni
Giovanni Millo, PhD
Research Dept.,
Assicurazioni Generali SpA
Via Machiavelli 4,
34132 Trieste (Italy)
tel. +39 040 671184 <tel:%2B39%20040%20671184>
fax +39 040 671160 <tel:%2B39%20040%20671160>
--------------- original message ---------------
Message: 18
Date: Thu, 26 Apr 2012 14:07:16 +0200
From: Ruben de Bliek <rubendebl...@gmail.com>
To: r-help@r-project.org
Subject: [R] PLM package PGGLS strange behavior
Message-ID:
<CAMjFNLVpiDsSVemYf=ctx2fzljqekhoobnctc8fc2csrgkx...@mail.gmail.com>
Content-Type: text/plain
When using the PLM package (version 1.2-8), I encounter the probem that
calling the FGLS estimator evokes strange behavior, when choosing the
"random" effects model. After calling the PGGLS function to estimate
FGLS,
PLM gives me a warning, stating that the "random" model has been
replaced
with the "pooling" model. I would, however, really like to estimate the
random model instead. For me, the problem is reproducable using one of
the
examples from the PLM Jstatsoft article "Panel Data Econometrics in R:
The
plm package" (pp.19-20):
data("EmplUK", package="plm")
zz <- pggls(log(emp)~log(wage)+log(capital),data=EmplUK, model="random")
summary(zz)
Which for me results in the following warning:
WARNING: Warning: 'random' argument to pggls() has been renamed as
'pooling'
It then proceeds with estimating a pooled model. I've checked if PLM by
any
chance does produce the right coefficient estimates, but the numbers do
not
add up when compared to the estimates in the article. This problem
perists
for any dataset I use. Any thoughts?
------------------ end original message ---------------------
Ai sensi del D.Lgs. 196/2003 si precisa che le informazioni contenute
in questo messaggio sono riservate ed a uso esclusivo del destinatario. Qualora
il messaggio in parola Le fosse pervenuto per errore, La invitiamo ad
eliminarlo senza copiarlo e a non inoltrarlo a terzi, dandocene gentilmente
comunicazione. Grazie.
Pursuant to Legislative Decree No. 196/2003, you are hereby informed
that this message contains confidential information intended only for the use
of the addressee. If you are not the addressee, and have received this message
by mistake, please delete it and immediately notify us. You may not copy or
disseminate this message to anyone. Thank you.
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
