Unit deviation vs. standard deviation shocks

[trams]

Hello everyone,

Could someone explain to me why the authors are doing the following to get their shocks and impulse responses?

This paper estimates a factor-augmented VAR. The unobservable factors are estimated from a standardised dataset. The unobserved factors and observed factors (variables that are non standardised) are then estimated together in a VAR.

Each variable is shocked to obtain the impulse responses. However, the way the shocks have been implemented are different.

All shocks are normalised to one unit deviation. For shocks to the observable factors, the authors have adjusted the normalisation by the standard deviation of the actual variable. For other variables except for the observable factors (which would refer to the unobserved factors), the impulse responses are displayed in standard deviation units.


So, what I understand from the above is that impulse responses for the unobserved factors would be in standard deviation units (since they are from a standardised dataset), while impulse responses for the observed factors would just be in their original units. What I don't understand is why is there a need for shocks to the observed factors to be adjusted by the standard deviation of the actual variable, while shocks to the unobserved factors are just a one unit deviation?

Would appreciate any help. Thanks!