With the adoption of IFRS 9, the debtor provision was reset to 22% of the book (compared to 9.5% in FY18). We show that TFG’s previously used Markov Model was highly predictive of actual bad debt write offs, and we argue that over an extended 13-year period from 2004 to 2017, an accurate forward-looking expected credit loss-based provision should approximate the result of the Markov Model. We therefore believe TFG’s level of provisioning may be overly conservative.
