What approach needs more data, holdings or transaction-based attribution? (Part II)

Earlier this week I mentioned that a colleague told me that transaction-based attribution requires a lot more data, which opens it up to the risk of errors being introduced. Two quick responses arise:

1) Is it true that transaction-based attribution requires more data than holdings?
2) How dirty is the data?

We discussed the first; now briefly the second.

Some have argued that given the increased volume of data needed to support transaction based attribution, you'd have a greater likelihood of data errors, which could corrupt your report; therefore, you should stick with holdings-based, which, following this logic, has less data, therefore has a lower likelihood of dirty data.

As was pointed out on Tuesday, the first point (that transaction-based attribution requires more data) seems to be invalid. But let's still consider the data's integrity.

Most asset managers reconcile their portfolios regularly with their official books and records. We also know that if transactions are "dirty," they won't settle. Yes, we know errors can and do occur, but they're typically corrected. I see this as a spurious argument, too.

As will be pointed out in my dissertation, there is significant justification for transaction-based attribution; but, you'll have to wait for the movie, based on the dissertation (kidding, of course; an article) to get the details behind this claim.