oo_.SmoothedShocks

Estimation ML/Bayesian estimation
1

Hi everyone,

I have two questions:

  1. I am working with an estimated DSGE model where I want to perform some counterfactual analysis. I have read the manual several times trying to understand how I should interpret the shocks in oo_.SmoothedShocks, but it is not clear to me. Since (if I set the state at the first observation in oo_.SmoothedVariables) and simulate the model with the smoothed shocks I get back my observable series, can I interpret these shocks as the shocks necessary (based on my model) to generate the observed series which I use in the estimation?

  2. After estimating my model, are the matrices for transitions in oo_.dr based on the posterior means or modes (I am using mh_replic>0)?

Thanks,
J.

2
  1. Yes, the smoothed shocks are our best estimate of what shocks historically gave rise to the observed data.
  2. After a MCMC, the parameters will be set to the posterior mean. However, the decision rules were not yet recomputed, so they will be some stale value from before.