Dear Sayan, no, unfortunately I don't think it will. Here's basically how coeftest() works: if you call the coeftest() function on a model object, say: 'mymodel', it will apply both a 'coef' and a 'vcov' method to mymodel in order to extract beta and vcov(beta) and do a Wald test. coeftest() works with many different kinds of models, represented by 'lm', 'glm', 'plm' objects and so on, each containing a 'standard' covariance matrix, so that the default behaviour is just to extract this latter. Alternatively, you can supply a vcov method of your choice to coeftest() and have it do robust testing etc., but it will still have to be one that fits your kind of model. So if 'mymodel' is a plm object, then > coeftest(mymodel, vcov=vcovHC) will use the White-Arellano covariance matrix, which as observed is robust vs. serial correlation in its peculiar way, different from the Newey-West-based vcovHAC for 'lm' objects. I'm too ignorant of the subject to give advice on tobit models, but a quick glance (?tobit) reveals that 'tobit' class objects inherit from 'survreg' ones, so that's the direction in which to look. Maybe you are in a position to simply pool the data and use standard tobit and vcovHAC? Panel data would have N observations out of NT that are serially uncorrelated by construction, and of course this would imply the assumption of no individual effects whatsoever (but I am just guessing here...). Best wishes, Giovanni
________________________________ Da: sayan dasgupta [mailto:kitt...@gmail.com] Inviato: mercoledì 9 dicembre 2009 06:59 A: Millo Giovanni; Achim Zeileis; yves.croiss...@let.ish-lyon.cnrs.fr Cc: r-help@r-project.org Oggetto: Re: Serial Correlation in panel data regression Dear Sir, Thanks for your reply But still exists a trick . Basically I want to do Panel Tobit. I am using the tobit function from the package (AER) on a panel data . Suppose that Gasoline$lgaspcar is a 0 inflated data and I do m1<- tobit (as.formula(paste("lgaspcar ~", rhs)), data=Gasoline) then if I do library(lmtest) coeftest(m1,vcovHC) Will it take account of the heteroskedasticity and serial correlation( within country ) of the data Regards Sayan Dasgupta On Tue, Dec 8, 2009 at 8:29 PM, Millo Giovanni <giovanni_mi...@generali.com> wrote: Dear Sayan, there is a vcovHC method for panel models doing the White-Arellano covariance matrix, which is robust vs. heteroskedasticity *and* serial correlation, although in a different way from that of vcovHAC. You can supply it to coeftest as well, just as you did. The point is in estimating the model as a panel model in the first place. So this should do what you need: data("Gasoline", package="plm") Gasoline$f.year=as.factor(Gasoline$year) library(plm) rhs <- "-1 + f.year + lincomep+lrpmg+lcarpcap" pm1<- plm(as.formula(paste("lgaspcar ~", rhs)), data=Gasoline, model="pooling") library(lmtest) coeftest(pm1, vcov=vcovHC) Please refer to the package vignette for 'plm' to check what it does exactly. Let me know if there are any issues. Best, Giovanni -----Original Message----- From: Achim Zeileis [mailto:achim.zeil...@wu-wien.ac.at] Sent: Tue 08/12/2009 13.48 To: sayan dasgupta Cc: r-help@R-project.org; yves.croiss...@let.ish-lyon.cnrs.fr; Millo Giovanni Subject: Re: Serial Correlation in panel data regression On Tue, 8 Dec 2009, sayan dasgupta wrote: > Dear R users, > I have a question here > > library(AER) > library(plm) > library(sandwich) > ## take the following data > data("Gasoline", package="plm") > Gasoline$f.year=as.factor(Gasoline$year) > > Now I run the following regression > > rhs <- "-1 + f.year + lincomep+lrpmg+lcarpcap" > m1<- lm(as.formula(paste("lgaspcar ~", rhs)), data=Gasoline) > ###Now I want to find the autocorrelation,heteroskedasticity adjusted > standard errors as a part of coeftest > ### Basically I would like to take care of the within country serial > correlaion > > ###that is I want to do > coeftest(m1, vcov=function(x) vcovHAC(x,order.by=...)) > > Please suggest what should be the argument of order.by and whether that will > give me the desired result Currently, the default vcovHAC() method just implements the time series case. A generalization to panel data is not yet available. Maybe Yves and Giovanni (authors of "plm") have done something in that direction... sorry, Z 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.
