I presume that your problem is in quantitative macroeconomics and
that your sample size is limited. Are your variables stationary. If
not you may need to use a VECM or if there is no cointegration work in
first differences.
My choice of variables would in the first instance be determined by
economic theory and the number of lags by an information criterion or
by testing down from a general model. You might use Granger Causality
tests to reduce the model if necessary. Otherwise a lot depends on
the properties of your model.
For forecasting purposes a BVAR might be a better solution.
Best Regards
John Frain
2008/8/12 Zhang Yanwei - Princeton-MRAm <YZhang at
munichreamerica.com>:> Hi all,
> I got another VAR question here and really appreciate if somebody would
help me out :)
> I have five time series, say A,B,C,D,E. My objective is to predict the
series A using the rest, that is, B, C, D and E. A Vector Autoregression Model
should work here. But first of all, I should select which series of B, C, D and
E to be include in the VAR model, as well as the number of lags. I wonder If
someone would give some advice on such a series selection procedure. Thanks so
much.
>
> Sincerely,
> Yanwei Zhang
> Department of Actuarial Research and Modeling
> Munich Re America
> Tel: 609-275-2176
> Email: yzhang at munichreamerica.com<mailto:yzhang at
munichreamerica.com>
>
>
> [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list
> stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
--
John C Frain
Trinity College Dublin
Dublin 2
Ireland
tcd.ie/Economics/staff/frainj/home.html
mailto:frainj at tcd.ie
mailto:frainj at gmail.com