The RATS Software Forum

Just a quick query: I am using montevar to do impulse responses using cholesky decomposition. Shouldn't the impulse response on impact for the last variable due to the last shock be unity? That is, aren't we doing impulse responses due to a 1 s.d. shock?

Luching.

luching wrote:Just a quick query: I am using montevar to do impulse responses using cholesky decomposition. Shouldn't the impulse response on impact for the last variable due to the last shock be unity? That is, aren't we doing impulse responses due to a 1 s.d. shock?

Luching.

No. You're correct that they're one s.d. shocks (to the orthogonalized residuals). But the responses are measured in the natural scale of the variables. A .03 impact for a variable measured in logs means 3%, a .25 response for an interest rate means a quarter of a point.

Thanks. I found out my impulse responses normalized the s.d. of shocks to be unity.