I am estimating a model and receiving the following message:
“DiffuseLiklihood3 :: T does influence the rank of Pinf!”

This occurs during the optimization routine (I think) when evaluating the likelihood function.

Is there something obvious that might explain why this error occurs?

I am using bayesian estimation for a model with non-stationary processes (so specified in the options_.unit_root_vars), though I am not using the bvar-dsge framework. I have not specified observations_trends – Is this required for the unit-root processes?

Thanks in advance.

Could you send us the program and the data?

It is something that may happen with the diffuse Kalman filter, but I need the example to see exactly what to do.

You should set observation_trends only if there is a deterministic trend component in your model