Should We Trust Your RWA Models?

According to today’s Financial Times, 63% of investors have lost faith in bank’s internal models to measure their risk-weighted assets (RWA). In an article titled “Investors lose faith in banks’ RWA models,” the paper says that many investors surveyed recently by Barclays would like to see big banks use the “standardized” method of measuring RWAs for Basel, rather than their own models as allowed under the internal ratings based (IRB) approach.

The Basel Committee established the IRB so that banks could use their knowledge and experience to move away from the Basel I “one size fits all” approach. Many have invested in large modelling teams that develop models that are more accurate for their own portfolios. Sceptics feel  banks may be “fudging” those models to reduce their reserve requirements — in other words, making them less accurate, not more.

FICO’s not a regulator, and we wouldn’t presume to comment on what regulators or institutional investors feel may be happening with internal bank models. But we do feel the IRB is worth preserving.

The wisest banks use their RWA models not just to calculate capital requirements but to make day-to-day lending decisions.  These banks have every incentive to make these models as robust and accurate as possible. If critics doubt the integrity of these models, more scrutiny should be given to the methodology used, the changes that result in adjustments to key metrics like loss given default, and the validation process. This isn’t rocket science — it’s what regulators worldwide do to ensure the safety and soundness of lending practices.

FICO has worked with many clients in EMEA and elsewhere on Basel model development and validation. We have helped lenders adopt economic impact models that offer a better way to project the effects of different economic cycles on credit risk in a consistent and explicit way. I have personally worked with clients to make their RWA calculations stronger, which may indeed reduce capital requirements, but only because the standardized approach — being less precise — overestimates the risk for many portfolios.  If this frees up more capital, it will also help lenders meet that other demand of regulators — putting more credit into the local economy.

Internal models should be more precise, and doubts about their validity should be laid to rest with facts, not the elimination of a good practice. We stand by the intent of the IRB.