Data with different frequency

#1

Dear,

I’m about to estimate a DSGE model, but one of my variables is only available on annual basis. My question is : if I change the frequency of that variable from annual to quarterly (using interpolation) will this affect badly my results ?

Thank you

#2

That is not easy to answer. Using interpolation you are not adding information to your data. In this case, would rather go for a mixed-frequency estimation approach.

#3

Thank you.
Is Dynare can perform a Bayesian Mixed Frequency Estimation ?

#4

Yes, see the respective section in Pfeifer(2013): “A Guide to Specifying Observation Equations for the Estimation of DSGE Models”

#5

Thank you again for the reply,
I read the section dealing with mixed frequency, and now I know how to perform MFE in Dynare.
However, I didn’t found enough technical details on how Dynare performs this. is there any paper which will help me to go a little bit deeper in that direction ?

#6

Hi Pfeifer,
Eq (70) Y_q1= Y_m1+ Y_m2+ Y_m3 in Pfeifer(2013): “A Guide to Specifying Observation Equations for the Estimation of DSGE Models”. If I detrend the Y observed data (percentage deviation from ss) to obtain Y_obs here. How to build the observed equation between Y_obs and Y_q1 here. Because Y_q1_hat is not equal to Y_m1_hat+ Y_m2_hat+ Y_m3_hat

Thanks!

#7

In logs, this would be the average. See e.g. the Appendix to Born/Pfeifer (2014): “Risk Matters: A comment”, AER