Observed variables

Estimation ML/Bayesian estimation
1

Hi
I’m new using dynare. I’m trying to stimate a model with 84 variables and 84 equations. In addition I have three exogenous observed variables. So, I followed the recomendations of Pfeifer’s guide adding the name of the variables in the list of endogenous variables, then the same names with “_obs”. I added three equantions in the model section and before the estimation command I added the observed variables with the command varobs. Finally, I added the data in the stimation command. The error message says 87 equations and 90 endogenous variables. I’m using a xls file I don’t know if the problem is the data or something else.

thanks

2

I would need to see the files.

3

At this point the code has many issues that I’m trying to fix. But I need to fix this first.
Here the code and data.base_ss_v2.xls (26.5 KB) ss_v2.mod (12.9 KB)

4

I am not following what you are doing in your code. You have three identities

p_oil_obs = p_oil; 
v_obs = v;	
ap_obs = ap;

defining three variables. But that still leaves 87 other variables with 84 equations while you stated that you have 84 variables in 84 equations.

5

I think I understand your point. If I add identities instead equantions I would have more endogenous variables than equations. However, in your guide (page 16) you used an indentity for the interest rate (R_obs=R), so I get confused.

So the solution could be to add a kind of shock or a process in those aditional equations and then specify the shock?

6

Before moving to estimation, you need a model that runs with simulations. Only then you should add the equations defining the observables.

7

But if I have 84 variables and 84 equations, and three aditional observable variables how would it runs without any information about the observable variables. Can I treat them as exogenous variables in varexo?
After that, if the simulation runs, How can I specify the equations to treat them as observable variables?.

8

For the three additional observable variables there would be three equations defining them.

9

In your previous answer you said that the model must run berofe defining the equations for this variables. But if I can’t use data to treat them as observable variables to make the model run, can I avoid _obs and treat them as a exogenous variables with their respective equations?

10

If those variables are exogenous, then you should define them as such.

11

Thank you.
Last question.
The examples that I already have seen about varexo refers to shocks. Can you recomend me a guide or something else about exogenous variables.

12

In the context of stochastic simulations with their stationary setup, exogenous variables are usually shocks. It’s different for perfect foresight simulations. I am not aware of any guides in this respect.