R help - Jan 2013 - plm random effect: the estimated variance of the individual effect is negative

Matteo,

I fully agree with David: please read the posting guide.

Anyway, the error message says it all: "the estimated variance of the
individual effect is negative". See e.g. the "basic panel"
chapter (10
or 11) in Wooldridge's "Econometric Analysis of XS and Panel Data"
to
understand why this may happen.  

Stata's behaviour is (as far as I remember) to transparently substitute
it with 0; 'plm's is to let the estimation fail, issuing a warning. You
can try a different 'random.method' which may, or might not, solve the
problem, depending on your data. 

Hint: a random effects model with sigma_mu=0 is a pooling model.

HTH

Giovanni Millo, PhD
Research Dept.,
Assicurazioni Generali SpA
Via Machiavelli 3,
34132 Trieste (Italy)
tel. +39 040 671184
fax  +39 040 671160

------------ original message ------------------

Message: 49
Date: Thu, 3 Jan 2013 13:23:07 -0800
From: David Winsemius <dwinsemius at comcast.net>
To: matteo ognibene <ognibenematteo at hotmail.it>
Cc: "r-help at r-project.org" <r-help at r-project.org>
Subject: Re: [R] plm random effect: the estimated variance of the
	individual	effect is negative
Message-ID: <D32FDC11-38AE-4AAF-9270-30C791232A24 at comcast.net>
Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes


On Jan 3, 2013, at 8:19 AM, matteo ognibene wrote:
>
> Hi,R 2.15.2 plm() function on Windows 7
> I have a problem with the variance estimation in a random effect model
> I used this formula to get my result, and it works:reg <- deltaF ~  
> L1.deltaF + L2.deltaF + deltaCDS + L1.deltaCDS + L2.deltaCDS +  
> L3.deltaCDS + deltaUS_Yields + deltaZ + L1.deltaZ + L2.deltaZfixed  
> <- plm(reg, data=Mody_R, na.action = na.exclude,  
> index=c("country_id", "datestata"),
model="random")
> but when i split my dataframe into several groups it gives me this  
> error:"Error in swar(object, data, effect) :   the estimated  
> variance of the individual effect is negative"
> The code i used to split the dataframe is:SplitMody <- split(Mody_R,  
> Mody_R$period)SplitFixed <- lapply(SplitMody, function(Mody_R)  
> plm(reg, data=Mody_R, na.action = na.exclude,
index=c("country_id",
> "datestata"), model="random"))
> I want to point the all the code works perfectly in a within model,  
> and I tried using all the random.method for the variance  
> estimation.In addition I used the same model in Stata and it  
> works... (the results are identical for a within model in R and Stata)
> I'll be really gratefull if someone can help me.
> Thanks,Matteo
> 	[[alternative HTML version deleted]]
Head up; matteo: This is pretty much unreadable (due to the failure to  
follow the recommended posting procedure.)

-- 
David Winsemius, MD
Alameda, CA, USA



--------------- end original message ---------------

?
Ai sensi del D.Lgs. 196/2003 si precisa che le informazi...{{dropped:12}}

Thanks Giovanni,

but I already tried with the others "random.method" and it doesn't
work anyway.
Do you think is it possible change the if statement in the plm formula, to set
the value to 0 if sigma<0 ?
Because if I insert a dummy variabile in the formula R gives me this error:
> Error in if (sigma2$id < 0) stop(paste("the estimated variance of
the", ?:?
> ? ? ? ? ? ? missing value where TRUE/FALSE needed
Thanks
Matteo Ognibene


----------------------------------------
> Subject: Re: [R] plm random effect: the estimated variance of the
individual effect is negative
> Date: Fri, 4 Jan 2013 16:26:01 +0100
> From: Giovanni_Millo at Generali.com
> To: ognibenematteo at hotmail.it
> CC: r-help at r-project.org
>
> Matteo,
>
> I fully agree with David: please read the posting guide.
>
> Anyway, the error message says it all: "the estimated variance of the
> individual effect is negative". See e.g. the "basic panel"
chapter (10
> or 11) in Wooldridge's "Econometric Analysis of XS and Panel
Data" to
> understand why this may happen.
>
> Stata's behaviour is (as far as I remember) to transparently substitute
> it with 0; 'plm's is to let the estimation fail, issuing a warning.
You
> can try a different 'random.method' which may, or might not, solve
the
> problem, depending on your data.
>
> Hint: a random effects model with sigma_mu=0 is a pooling model.
>
> HTH
>
> Giovanni Millo, PhD
> Research Dept.,
> Assicurazioni Generali SpA
> Via Machiavelli 3,
> 34132 Trieste (Italy)
> tel. +39 040 671184
> fax +39 040 671160
>
> ------------ original message ------------------
>
> Message: 49
> Date: Thu, 3 Jan 2013 13:23:07 -0800
> From: David Winsemius <dwinsemius at comcast.net>
> To: matteo ognibene <ognibenematteo at hotmail.it>
> Cc: "r-help at r-project.org" <r-help at r-project.org>
> Subject: Re: [R] plm random effect: the estimated variance of the
> individual effect is negative
> Message-ID: <D32FDC11-38AE-4AAF-9270-30C791232A24 at comcast.net>
> Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes
>
>
> On Jan 3, 2013, at 8:19 AM, matteo ognibene wrote:
>
> >
> > I have a problem with the variance estimation in a random effect model
> > I used this formula to get my result, and it works:reg <- deltaF ~
> > L1.deltaF + L2.deltaF + deltaCDS + L1.deltaCDS + L2.deltaCDS +
> > L3.deltaCDS + deltaUS_Yields + deltaZ + L1.deltaZ + L2.deltaZfixed
> > <- plm(reg, data=Mody_R, na.action = na.exclude,
> > index=c("country_id", "datestata"),
model="random")
> > but when i split my dataframe into several groups it gives me this
> > error:"Error in swar(object, data, effect) : the estimated
> > variance of the individual effect is negative"
> > The code i used to split the dataframe is:SplitMody <-
split(Mody_R,
> > Mody_R$period)SplitFixed <- lapply(SplitMody, function(Mody_R)
> > plm(reg, data=Mody_R, na.action = na.exclude,
index=c("country_id",
> > "datestata"), model="random"))
> > I want to point the all the code works perfectly in a within model,
> > and I tried using all the random.method for the variance
> > estimation.In addition I used the same model in Stata and it
> > works... (the results are identical for a within model in R and Stata)
> > I'll be really gratefull if someone can help me.
> > Thanks,Matteo
>
> > [[alternative HTML version deleted]]
>
> Head up; matteo: This is pretty much unreadable (due to the failure to
> follow the recommended posting procedure.)
>
> --
> David Winsemius, MD
> Alameda, CA, USA
>
>
>
> --------------- 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.