The RATS Software Forum

Could you explain this?

bootstrap/[nobootstrap]
BOOTSTRAP draws shocks over the forecast period randomly with replacement
from the residuals associated with the equation or regression being forecast.
Because this is a simple shuffling of the residuals, it would not be completely appropriate
for a model with moving average terms if you’re bootstrapping an entire
sample.

I am a bit hung up on "not completely appropriate."

Ahh...MA model errors are correlated, but this assumes iid errors?

Correct. (This is for the BOOTSTRAP option on UFORECAST). The BOOT instruction has options for block bootstrapping which can handle some forms of weak serial correlation. However, parametric bootstrapping for an ARMA model is fairly complicated. See the BOOTARMA.RPF example.