This syntax can only be used with `perfect_foresight_with_expectations_errors\_\*` commands
### Extensions
#### Local approximation of the solution
If the model is linear or we already have the local approximation of a non-linear model, we should be able to use the above syntax while simulating with the approximated solution (requires new simulation function)
#### Information about future endogenous variables
We should have a similar syntax for providing information about future values of endogenous variable. In general, it requires to also specify the exogenous variables that need to be freed to solve the model. In the case of linear model, is also possible to use the Kalman filter (or smoother).
We can simply authorize `var` to design an exogenous OR AN ENDOGENOUS variable (maybe a bit confusing) or we can have separate blocks for exogenous variables (shocks) and another block name for future value of endogenous variables (may be too much writing)
#### Stochastic simulations
This scenario description could be embedded in stochastic simulations like extended path, repeating the simulation while drawing random values for the shocks not constrained by a piece of future information.
#### Future information about stochastic shocks
In the same manner we should allow for specifying new variance for the shocks in the future.
Currently in unstable we have
```
shocks(learnt_in=2);
var x;
periods 3:4;
add 0.1;
end;
```
This syntax can only be used with `perfect_foresight_with_expectations_errors\_\*` commands
### Extensions
#### Local approximation of the solution
If the model is linear or we already have the local approximation of a non-linear model, we should be able to use the above syntax while simulating with the approximated solution (requires new simulation function)
#### Information about future endogenous variables
We should have a similar syntax for providing information about future values of endogenous variable. In general, it requires to also specify the exogenous variables that need to be freed to solve the model. In the case of linear model, is also possible to use the Kalman filter (or smoother).
We can simply authorize `var` to design an exogenous OR AN ENDOGENOUS variable (maybe a bit confusing) or we can have separate blocks for exogenous variables (shocks) and another block name for future value of endogenous variables (may be too much writing)
JP (2025/10/07): The `perfect_foresight_controlled_paths` provides the necessary syntax. We should mirror it for the stochastic case.
#### Stochastic simulations
This scenario description could be embedded in stochastic simulations like extended path, repeating the simulation while drawing random values for the shocks not constrained by a piece of future information.
#### Future information about stochastic shocks
In the same manner we should allow for specifying new variance for the shocks in the future.
Of course, we can´t and don't need to implement everything at once, but we should plan the syntax to make it possible
