Order of variables in multi-variate GARCH models

[sanjeev]

I am using VAR(2)-BEKK(1,1) model to estimate shock and volatility spillovers among three exchange rate returns series.I noticed that my results differ with the order in which i enter the variables in the command.I guess its something to do with the maximum likelihood procedure/BFGS method used for estimating BEKK model.Kindly explain the reasoning behind such results.


Thanks,
Sanjeev

[TomDoan]

If you re-order the variables, all the "labeling" will change. The log likelihood should be the same, however. There are only a handful of GARCH models where the ordering affects the fit (triangular BEKK and Cholesky).