Residual autocorrelation in Bayesian VAR

[ntchuong]

Hi guys,

I conduct Bayesian VAR model to make conditional forecast. However, VAR Residual Portmanteau Tests for Autocorrelations shows that the residuals is correlated.
I tried to add more variables and change the lag order, but it doesn't work.
Can you please tell me how to overcome this problem?
Many thanks in advance.

[PoachedWonk]

I am having a similar issue, It seems that by extending the lags, autocorrelation becomes more of an issue, rather than less...

Can I use the Portmanteau / AC LM test on BVARs? Or am I doing something wrong??

[PoachedWonk]

Hi the resolution to this is to change the priors, if lambda 1 is too low, then we assume the prior values are more or less correct, which restricts the convergence of the model and leads to autocorrelation.