On Sat, 30 Mar 2013, N. Janz wrote:
> Dear all,
>
> how can I use vcovHC() to get robust/corrected standard errors from an
> arima() output?
I'm not sure whether it is feasible to apply HC corrections to ML
estimates of ARIMA models. It's not clear to me whether these would really
be robust against heteroskedasticity of the innovations.
> I ran an arima model with AR(1) and got the estimate, se, zvalue and
p-value
> using coeftest(arima.output).
>
> However, I cannot use vcovHC(arima.output) to get corrected standard
> errors. It seems vcovHC works only with lm and plm objects?
In principle, vcovHC() or sandwich() can be applied to a wider range of
models that provide certain methods: Either a vcovHC method directly (like
plm) or an estfun method for sandwich etc. For details see
vignette("sandwich-OOP", package = "sandwich")
Whether it is sensible to apply these method (or whether it makes the
standard errors robust against anything) is a different matter though.
Best,
Z
> Is there another way I can get robust/corrected standard errors, or am I
> missing something?
>
>
>
> Thank you!
>
>
> Nicole
>
>
> --
>
>
> I got this error:
>
>
>
>> coeftest(arima.res.total,vcovHC)
>
>
>
> Error in terms.default(object) : no terms component nor attribute
>
>
>
>
> I also tried this:
>
>
>
>> coeftest(arima.res.total, vcovHC=vcovHC(arima.res.total,
>> method="arellano"))
>
>
>
> I do get an output table, but the standard errors do not change at all from
> the original coeftest() table, so I'm not sure it did the job.
>
>
>
>
> -
> R version 2.15.2 (2012-10-26)
>
>
> Platform: x86_64-apple-darwin9.8.0/x86_64 (64-bit)
>
>
> locale:
>
>
> [1] C/en_US.UTF-8/C/C/C/C
>
> attached base packages:
>
>
> [1] grid splines stats graphics grDevices utils datasets methods base
>
> other attached packages:
>
> [1] ellipse_0.3-7 corrgram_1.4 seriation_1.0-10 colorspace_1.2-0
gclus_1.3.1
> TSP_1.0-7 cluster_1.14.3 car_2.0-15 nnet_7.3-5 [10] tseries_0.10-30
pcse_1.8
> arm_1.6-04 foreign_0.8-51 abind_1.4-0 R2WinBUGS_2.1-18 coda_0.16-1
> lme4_0.999999-0 Matrix_1.0-10 [19] lattice_0.20-10 gplots_2.11.0
> KernSmooth_2.23-8 caTools_1.14 gdata_2.12.0 gtools_2.7.0 Hmisc_3.10-1
> survival_2.37-2 simcf_0.2.8 [28] lmtest_0.9-30 plm_1.3-1 sandwich_2.2-9
> zoo_1.7-9 MASS_7.3-22 Formula_1.1-0 nlme_3.1-106 bdsmatrix_1.3
>
> loaded via a namespace (and not attached):
>
> [1] bitops_1.0-4.2 quadprog_1.5-4 stats4_2.15.2 tools_2.15.2
>
> ______________________________________________
> R-help at 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.
>