Skip to content

VECM estimation

4 messages · Jonathan Beokhokhei, John Kerpel, Pete B +1 more

#
Dear all, I am working with some VECM estimation methodology, however could not find any complete package for doing so in R. I am aware of?the "urca" package, however could not find any function for :

1. Availing "t-ratio" with estimated individual estimated SD, for all parameters like Alpha, Beta, Sigma and other model parameters, under different setup for deterministic terms like simple constant, deterministic trend and orthogonal deterministic trend, with varying lag and rank parameters.

2. Performing residual analysis like plotting ACF, PACF for estimated residuals with confidence band and portmanteau testing etc.

Ofcourse I can write for my own, however it obviously not a good idea to reinventing the wheel. Is there any functions available for doing above?

Thank you for your time.
1 day later
#
Jonathan

Combination of package vars and urca will provide what you need, showing 
different p-values of the lags and alpha coefficients, including 
deterministic regressor (but no orthogonal one), and various 
specification tests. P-values for beta are, as far as I know, not 
available, but function blrtest() from urca can test for the betas.

Package tsDyn, which deals with nonlinear VAR and VECM, has a few 
alternative VAR and VECM() estimation and representation functions, with 
a toLatex() method to export the VAR/VECM matrix equation into Latex.

Hope this helps

Matthieu



Le 09. 10. 10 18:07, Pete B a ?crit :