Question on Shock Ordering and Identification in DSGE Models

Dear Dynare Community,

I have a question regarding the treatment of shocks in DSGE models, specifically about the importance of shock ordering and identification.

If I have five shocks in a DSGE model, and only two of them are correlated while the others are uncorrelated, does the order of these correlated shocks matter relative to the other three uncorrelated shocks? I understand from the Dynare manual that the ordering of correlated shocks is important, similar to VAR models. However, does this order also influence how Dynare processes shocks when some are uncorrelated?

Additionally, does a DSGE model require shock identification, like in a VAR setting, or are all shocks predefined without needing further identification procedures? Could you help clarify how Dynare handles these shocks internally and how the process differs from shock identification in VAR models?

Thank you for your guidance and insights!

Best regards,

Yamoi

  1. DSGE models by construction start from structural shocks, i.e. theses structural shocks are already identified shocks. You don’t need to back out the structural shocks from the reduced form ones you would get in a VAR.
  2. The uncorrelated shocks in a DSGE model can be order wherever you want. They will not affected by the correlated ones. Note also that the ordering within the block of correlated shocks only matters for the purpose of generating IRFs and variance decompositions. Here, you need to uniquely assign effects to particular shocks, i.e. you need to orthogonalize. It will not matter for the overall model fit.

Thank you! That’s very helpful!