Requirements for choosing the number of observed variables

Dear Prof Pfeifer,
My model is better identified (and more stable) for example with six observed variables.
When I increase the number of these variables, identification and stability deteriorates.
It is normal?
What are the requirements for choosing the number of observed variables(except for the connection with shocks)?

Thanks a lot.

See the discussion of the topic in “Remark 7 (Which Observables for Estimation?)” of Pfeifer(2013): “A Guide to Specifying Observation Equations for the Estimation of DSGE Models