On Thu, 16 Mar 2017, alfonso.carf...@uniparthenope.it wrote:
Hi all,
I want to ask you which is the difference between the specifyng and not
specifyng the covariance matrix of the estimated coefficients when
performing the coeftest command.
coeftest(object, ...) computes Wald statistics for all coefficients. Hence
coef(object) is used to extract the coefficients and then, by default,
vcov(object) is used to extract the variance-covariance matrix. For lm()
models this computes the "usual" covariance matrix estimate assuming
homoskedastic and uncorelated errors.
When you supply coeftest(object, vcov = vcovHC) then a
heteroscedasticity-consistent covariance matrix estimate is used (HC3 by
default).
See vignette("sandwich", package = "sandwich") for more details.
I'm estimating a VECM model and I want to test the significance of the
short-run casual effects of the explanatory variables:
mod<-cajorls(ca.jo(data[,4:6], ecdet = "const", type="eigen", K=2,
spec="longrun"))$rlm
The command:
coeftest(mod)
give me different results with respect to this one:
V<-vcovHC(mod)
coeftest(mod,V)
______________________________________________
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
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.
