We need to talk about Collins

Standardised capital has become the binding constraint for all US G-Sibs bar Goldman and BNY Mellon

In the post-crisis years, regulators tried to thrash out a new deal on capital modelling – one that balanced the goals of risk-sensitivity, simplicity and comparability.

It wasn’t easy, because some regulators wanted to prioritise simplicity or comparability, while others (along with much of the industry) emphasised the importance of risk sensitivity. In the end, regulators agreed to kill off the least-robust bits – op risk, credit valuation adjustment and certain inputs for low-default loans – while adding guardrails elsewhere, in the form of add-ons and tougher approval regimes.

In theory, this would allow capital to be modelled where those models could be trusted, and preserve the vision of the pre-crisis capital regime – the idea that the system would be safer if dynamic, risk-based capital encouraged banks to make sensible business choices.

That vision is now being blotted out – in the US, at least.

This is a result of the Collins Amendment to the Dodd-Frank Act – named after the Republican senator for Maine, Susan Collins – which requires modelling banks to also calculate their capital adequacy using the regulators’ standardised methodologies, and to apply the tougher of the two.

After its introduction in 2013, the new framework seemed to be working as designed, but by the second quarter of 2015, Wells Fargo had dropped below the so-called Collins floor. Three more banks would follow suit in the second quarter of 2017, and one more did so the following quarter.

A Risk Quantum analysis across the eight systemically important US banks shows that by June 2018, Bank of America, Citi, JP Morgan, Morgan Stanley, State Street and Wells Fargo had all hit the Collins floor. What was intended as a backstop to modelled capital requirements has become the binding standard for the world’s largest banks.

Now Goldman Sachs is also about to hit the floor. Its total modelled RWAs currently sit just $10 billion above standardised RWAs. That’s down from $62 billion a year ago.  

Why is this happening? One likely driver is the shake-up of banks’ portfolios in the past five years. The industry has offloaded complex assets that attracted higher capital requirements under the advanced approaches, and their slimmed down, vanilla books now look much less risky to their internal models. Another could be that banks’ models and assumptions are drifting further from Basel’s own methodologies.

However, the current status of the US capital regime raises important questions for regulators and policymakers alike: if there is no capital relief on offer, what incentives do banks have to keep using and investing in their models? Are banks’ models underestimating risk, or are the standardised approaches exaggerating them?