This is because I think that for the shocks that are too severe and disastrous, we really need to specify the magnitude of these shock which must be larger than one-standard-deviation change in the error term as usually assumed in STATA.