Mixed frequency estimation

Greetings Johannes et al.

Background: monthly DSGE model that I’d like to estimate with some monthly data, but quarterly GDP. Model is log-linearized.

Question: for my additional equation, following the guide, in levels I’d have to specify

Y(Q, t) = Y(M, t) + Y(M, t+1) + Y(M, t+2),

but since the model is log-linearized, this would have to be input in the model block as

yhat(Q, t) = (1/3)*[yhat(M, t) + yhat(M, t+1) + yhat(M, t+2)],

where Y is GDP in level, yhat in percentage deviation from trend, Q is quarterly, and M is monthly. Is this adequate?

yhat in the model corresponds to yhat(M), while yhat(Q) corresponds to the yhat in my data file. In the above linearization, I used

Y(Q, ss) = 3*Y(M, ss)

as the steady-state equivalent. For some reason I am not 100% certain I have this correct, which is why I’m posting this question. Any help, as always, is appreciated!

Yes, this is correct. In your linearization, there appears the ratio of monthly to quarterly GDP on the right hand side, which should be equal to 1/3.

Okay, great. Thanks, Johannes.