The RATS Software Forum

Usually, we use the Monte Carlo simulation or Bootstrapping technique to compute the errors bands for point estimates. In this case, the point estimates are of course the average value of many point estimates from many different draws.
My question is that: in computing a VAR/VECM Decomposition of Forecast Variance, if I use the Monte Carlo simulation (or Bootstrapping) to generate only the average point estimates (but I don't need to report the standard errors), does this work make sense ? Or, what I'm doing is really stupid because nobody does like me ?

Best regards

There's nothing wrong with that. Many paper report the mean from bootstrapping or simulation as the "center" of the distribution. However, why would you go to that work and not also do bands.