Lütkepohl, Helmut Staszewska-Bystrova, Anna Winker, Peter
Year of Publication:
DIW Discussion Papers 1642
There is evidence that estimates of long-run impulse responses of structural vector autoregressive (VAR) models based on long-run identifying restrictions may not be very accurate. This finding suggests that using short-run identifying restrictions may be preferable. We compare structural VAR impulse response estimates based on long-run and short-run identifying restrictions and find that long-run identifying restrictions can result in much more precise estimates for the structural impulse responses than restrictions on the impact effects of the shocks.
impulse responses structural vector autoregressive model long-run multipliers short-run multipliers