Peer data can be useful for de novo institutions, but should no longer be used when institutions have a substantive amount of data on their portfolio to perform a historical loss methodology.