Q 27: The agencies seek commenters’ perspectives on other loss types for which the boundary between credit and operational risk should be evaluated further (for example, with respect to losses on HELOCs).
Q 28: The agencies generally seek comment on the proposed treatment of the boundaries between credit, operational, and market risk.
Q 29: The agencies seek comment on this approach to tranched guarantees on retail exposures and on alternative approaches that could more appropriately reflect the risk mitigating effect of such guarantees while addressing the agencies’ concerns about counterparty credit risk and correlation between the credit quality of an obligor and a guarantor.
The following areas are worthy of consideration regarding the boundary between credit and operational risk:
ATM/Credit/Debit card losses: fraud losses are highly predictable given transaction volume and past historical trending.
Check fraud losses: can be predictable based upon historical experience and transaction volume. There are some frauds that are unique or one-offs from a trending perspective, but possibly the outliers could be removed when considering future losses.
Overdraft (fraud) losses: some consistency exists based upon volume of overdrafts when trended over a period of time using historical data. Outliers would need to be excluded from the calculations.
Small business loan fraud: may also lend itself to trending and predicting losses.
However, firms should be responsible for ensuring that there is appropriate allocation of risks for credit and operational risk; this is not an area where additional ex ante guidance is required, as highlighted below (see question 28).
The NPR text aligns with our intended approach. However, credit risk losses that arise from an operational loss event are captured within our operational loss data for management information purposes, but are clearly flagged to avoid double counting for regulatory capital purposes.
If documented, legally enforceable guarantees exist, we believe these should be incorporated into risk weighting calculations.