`var_expectation` and `aggregate` function clarification

Stochastic simulations Semistructural Modeling
1

1. Semi-structural model, VAR expectation part

According to the manual part 4.25.2, if I set the options of horizon = a:b and time_shift = tau as a = 0, b = 1 and tau = 0, the result I get would be

var_expectation(y) = \mathbb{E}_t y_t + \beta \mathbb{E}_t y_{t+1} with horizon = 0:1, time_shift = 0

My question is, would this be different with the following setup from your manual’s example in 4.25.2, where horizon is set as horizon = 1 instead of horizon = 0:1 to make the following computation?

var_expectation(y) = \mathbb{E}_t y_{t+1} with horizon = 1

If that’s the case, it must be that horizon = 1 is equal to horizon = 1:1 not horizon = 0:1, right?

My point is that it is a bit confusing for an user who are usually interested in looking for the expression of \mathbb{E}_t y_{t+1} as a benchmark case, while horizon = 1 and horizon = 0:1 seem to be equivalent(given that the default a=0 is assumed…)

Please correct me if I am wrong.

2. aggregate function and block_name issue

I am using cherrypick and aggregate function to add up all the blocks I constructed, however, I have not been successful to find a way to denote a block’s name before aggregation. However, what I find is that after using the aggregate function, the generated mod file automatically writes block_name = C: for every equation name. Could you tell me how to specify the block_name before aggregate works so that I can tell where each equation is from and avoid the DYNARE from automatically writing block_name=C: ?

Thank you.

2

Dear Seokil,

The documentation about the horizon option was incorrect (thanks for reporting the issue). I fixed it here, the change will be visible here:

https://www.dynare.org/manual-unstable/the-model-file.html#var_expectation_model

in a moment. If horizon is a scalar we do not compute an expected discounted sum from 0 to horizon.

Regarding the naming of the blocks, I would need your files to understand why you obtain C: as a name for (all?) the blocks. To set the name of the blocks I add a equation tag called blockname in the the orginal mod files. You can look at an example here:

Best,
Stéphane.

3

Dear Stephane,

Thank you for the correction and guidnace.
As a matter of fact, I am currently in charge of developing the FRB/US type model for a central bank as well.
I recently had a chance to get some technical assists from the ECB BASE modeling team and am using DYNARE as a main developing tool.

Accordingly, I cannot help but participate in the workshop that you are planning from here, Semi Structural models in Dynare

Please let me know when the plans go ahead.(kangseokil@bok.or.kr)

Best,
Seokil

1 Like