Ani98
September 28, 2021, 2:02pm
1
Hi everyone,
The usual way is to construct the variance from squared expected values. See for example
/* Replicates the IRFs at the stochastic steady/ergodic mean in the absence of shocks
* by Basu/Bundick (2017): "Uncertainty shocks in a model of effective demand",
* Econometrica, 85(3), pp. 937-958
*
* Notes:
* - Due to pruning, one cannot simply use the stochastic steady state as the starting point
* for simult_.m as one would do with the deterministic steady state in a linear model. The reason
* is the pruned state space where one would need to provide the first, second, and third order components
* of the stochastic steady state. Dynare does currently not yet support this. For this reason, the impulse
* period is simply appended to the simulation for computing the stochastic steady state
* - Basu/Bundick only use 200 periods of burn-in for computing the stochastic steady state. This is not sufficient
* for convergence as can be easily verified by plotting a longer simulation. Setting true_stochastic_steady_state_IRFs=1
* therefore uses a longer burn-in, leading to slightly different IRFs.
*/
/*
* Copyright (C) 2016-17 Benjamin Born and Johannes Pfeifer
*
* This is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
show original