Information set for the expectation variables with geometric discounting

To have a possibility to change the timing of an information set for the expectation variables with geometric discounting. At this stage it is hard coded at time t.

added Madysson label

added enhancement label

We need to check that the timing of the information set does not affect the reduced form parameters.

mentioned in commit preprocessor@221daeab

mentioned in issue preprocessor#70 (closed)

mentioned in commit 87b1d5bd

Done for VAR based expectations in fe63082c, I suppose we should do the same for PAC expectations... Note however that the default periods for VAR expectations (t) and PAC expectations (t-1) are not the same.

@stepan-a, I am not sure that's the right place, but I have a feeling my problem is connected to this issue. When running pac estimation with new Dynare version (4.8 July) I obtain strange results. I checked "test" and actually if you run dynare\tests\pac\var-6\example1.mod in 4.7 and 4.8, you'll see that parameters of pac_expectation variable are diferent depending on a version. In 4.7 I obtain:

> parameters h0_pacman_eq0_var_AUX_DIFF_39_lag_1 h0_pacman_eq0_var_y_lag_1 h0_pacman_eq0_var_AUX_DIFF_39_lag_2 h0_pacman_eq0_var_y_lag_2;
> 
> h0_pacman_eq0_var_AUX_DIFF_39_lag_1 = 0.0000000000000000;
> h0_pacman_eq0_var_y_lag_1 = 0.0000000000000000;
> h0_pacman_eq0_var_AUX_DIFF_39_lag_2 = 0.0000000000000000;
> h0_pacman_eq0_var_y_lag_2 = 0.0000000000000000;
> parameters h1_pacman_eq0_var_AUX_DIFF_39_lag_1 h1_pacman_eq0_var_y_lag_1 h1_pacman_eq0_var_AUX_DIFF_39_lag_2 h1_pacman_eq0_var_y_lag_2;
> 
> h1_pacman_eq0_var_AUX_DIFF_39_lag_1 = -0.0204386323060280;
> h1_pacman_eq0_var_y_lag_1 = -0.0090103235508248;
> h1_pacman_eq0_var_AUX_DIFF_39_lag_2 = -0.0026679701308692;
> h1_pacman_eq0_var_y_lag_2 = -0.0089540800364464;
> 
> parameters pacman_pac_growth_neutrality_correction;
> 
> pacman_pac_growth_neutrality_correction = 0.1853271645736974;
> 
> 

In 4.8:

parameters h0_pacman_eq0_constant h0_pacman_eq0_var_AUX_DIFF_39_lag_1 h0_pacman_eq0_var_y_lag_1 h0_pacman_eq0_var_AUX_DIFF_39_lag_2 h0_pacman_eq0_var_y_lag_2;

h0_pacman_eq0_constant = 0.0000000000000000;
h0_pacman_eq0_var_AUX_DIFF_39_lag_1 = 0.0000000000000000;
h0_pacman_eq0_var_y_lag_1 = 0.0000000000000000;
h0_pacman_eq0_var_AUX_DIFF_39_lag_2 = 0.0000000000000000;
h0_pacman_eq0_var_y_lag_2 = 0.0000000000000000;
parameters h1_pacman_eq0_constant h1_pacman_eq0_var_AUX_DIFF_39_lag_1 h1_pacman_eq0_var_y_lag_1 h1_pacman_eq0_var_AUX_DIFF_39_lag_2 h1_pacman_eq0_var_y_lag_2;

h1_pacman_eq0_constant = 0.4146728354263051;
h1_pacman_eq0_var_AUX_DIFF_39_lag_1 = 0.0000000000000000;
h1_pacman_eq0_var_y_lag_1 = 0.0000000000000000;
h1_pacman_eq0_var_AUX_DIFF_39_lag_2 = 0.0000000000000000;
h1_pacman_eq0_var_y_lag_2 = 0.0000000000000000;

parameters pacman_pac_growth_neutrality_correction;

pacman_pac_growth_neutrality_correction = 0.1853271645736974; 

Do you know why? Shall I add an option?

P.S. The test folder belongs to 4.8 but I think that it is the same example as in 4.7.

closed