Simple model. Unkown error!!

[Newbie]

Hello boys & girls,

I read through this topic and I tried to follow your indications.

The model with which I am playing is in log deviations from its steady state and I have:
1. de-trended my data(which is quarterly by the way) with the HP filter, Lambda = 1600
2. taken first order log differences of the de-trended data
3. de-meaned the first order log differences
Please see data set attached.

The error message is also attached.
When I run the estimation it basically iterates more than 200 times and afterwards says that there was a posterior optimization problem.

My questions:
1. is my data set ok? Is the model too big for only 5 observables?
2. are my measurement equations ok? Lines 460-464;
3. Next I would like to add more observables like inflation for example. I will consider year-on-year inflation, quarterly data. How would a measurement equation look like? for example obs_inflation = inflation - log(inflation_steady_state)?

Please help a desperate man! Good bless you!
Newbie.

[Ravelomanana]

My model variables is in deviation from the steady state so I manage my observable variables like this:
1- log transformation
2-Use HP filter to extract the deviation from the steady state
So My observation equation is y observable variable = y of the model. Dont need to make a transformation to match data to the model
am I right?

[jpfeifer]

@Newbie: Please see Pfeifer(2013): "A Guide to Specifying Observation Equations for the Estimation of DSGE Models" https://sites.google.com/site/pfeiferecon/Pfeifer_2013_Observation_Equations.pdf.

@Ravelomanana Do not use the two-sided HP-filter (see the linked document). Apart from that, you are correct. Your data and your model will both be percentage deviations from steady state/the BGP.