Autocorrelation in VAR model

[Atul Anand Jha]

I am running a vector autoregressive model of a dataset for 80 quarters. After running the commands varsoc the lag length is 2 and the varstable command specifies that the model is stable. Varlmar command for autocorrelation shows p value for both lag to be less than 0.05. Does this mean that there is autocorrelation among the residuals and if yes then how can I correct it? Or can I continue with the Granger Causality Test

[Justin Niakamal]

One approach is to allow for more lags. As for your second question, you'll want to add more lags until autocorrelation isn't an issue, then you can run a Granger Causality test. Dave Giles has an excellent post about this. I suggest you read:

https://davegiles.blogspot.com/2011/...causality.html