Measurement equations notation and deterministic trends

Topics: Kalman Filtering, Models
Jan 2, 2015 at 9:01 PM
Hello,

I am little bit confused about the notation of measurement equations and the inclusion of deterministic trends. Example: if the measurement variables for the consumption in my model is "c", and the deterministic component is a constant and linear time trend
 !dtrends
            dataC += dataC_  +dataCt_ * !ttrend;
should I write as a measurement equations for the consumption "c" in deviation from its steady-state
!measurement_equations
            'Observables - Consumption'
            dataC   = log(c/$c);
or in level
!measurement_equations
            'Observables - Consumption'
            dataC   = log(c);
In the estimate function, I set "deviation" to false and "dtrends" to true.

Thanks,
Gabriel