error bands

[MC128]

Hi all,

I have a question with regard to drawing the error bands of impulse responses. After estimation by ordinary least square, many papers used monte carlo simulation with several hundred replications to draw the 1 standard deviation error bands. I would like to ask whether there is any conceptual advantage or problem to draw 16-84 percentiles error bands instead. is the possibility of capturing asymmetric response one advantage?

Many thanks.

MC