What consensus? I've never heard this as a recommendation.
Well, I guess I was referring to the consensus of bankers and examiners I have talked to over the years. I have heard story after story that is consistent with what Moman said: They voluntarily resubmitted before their exam and their regulators never questioned the resubmission and also conducted their HMDA validation based off the resubmission (new) data.
Besides, once you're past the cut-off, fixing the errors is the equivalent of writing 100x on the blackboard. Your corrections are not reflected in the aggregate data.
A seasoned FDIC examiner once explained to me (which I later confirmed with the FRB) that resubmitted HMDA data would be corrected in the (nationwide) aggregate file up to about 2 1/2 years after the original submission date. While they would not update the bank's individual HMDA reports, the aggregate data is still updated - which is exactly why the regulators make banks resubmit HMDA data from a year or two ago.
In the past, HMDAHelp even gave me an exact date (about 2 1/2 years after the original submission deadline) for which resubmission data would no longer be added to the aggregate file, meaning that I had to get my voluntary resubmission in before that date.
My understanding is that the bottom line is that if you don't voluntarily resubmit, the examiners review the original data that has known errors and the bank risks a cited violation as well as a forced resubmission.