search

jmaih / RISE_toolbox

Solution and estimation of Markov Switching Rational Expectations / DSGE Models
BSD 3-Clause "New" or "Revised" License
106 stars 78 forks source link

Identification in IRF #162

Closed timqzhang closed 2 years ago

timqzhang commented 3 years ago

Dear Junior,

I notice a issue when running the tutorials in Bjornland case (exchange rate). If I run the IRF code (tut06 entirely) twice or more times with nothing changed, I will get some different results (sometimes different a lot). I use the same identification schemes. Is it due to some unstableness in the identification or IRF engine? Thank you.

Best, Tim

jmaih commented 3 years ago

There are lots of different impulse responses that satisfy sign restrictions.

In general people average across a huge number of simulations to get results that are approximately stable.

Cheers,

J.

On Sat, Mar 13, 2021 at 9:34 PM timqzhang @.***> wrote:

Dear Junior,

I notice a issue when running the tutorials in Bjornland case (exchange rate). If I run the IRF code (tut06 entirely) twice or more times with nothing changed, I will get some different results (sometimes different a lot). I use the same identification schemes. Is it due to some unstableness in the identification or IRF engine? Thank you.

Best, Tim

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/jmaih/RISE_toolbox/issues/162, or unsubscribe https://github.com/notifications/unsubscribe-auth/AATKBT4PQ7MRL2CEZZ4NFYLTDO45XANCNFSM4ZEFXYKQ .

timqzhang commented 3 years ago

Thank you for response. So I assume that for sign and long-run restrictions, currently RISE doesn't do the stableness part you mentioned to the identification and IRF?

jmaih commented 3 years ago

The answer is the one I gave you earlier. It is your responsibility to set the number of simulations

On Sat, Mar 13, 2021, 21:43 timqzhang @.***> wrote:

Thank you for response. So I assume that for sign and long-run restrictions, currently RISE doesn't do the stableness part you mentioned to the identification and IRF?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/jmaih/RISE_toolbox/issues/162#issuecomment-798782650, or unsubscribe https://github.com/notifications/unsubscribe-auth/AATKBT66OXHXN7ML3MT67Q3TDO54VANCNFSM4ZEFXYKQ .