Regulators are increasingly mistrustful of banks using internal models to calculate their capital requirements, but the alternatives are not trouble-free.
The Basel II international agreement brought with it the innovation that banks should hold capital according to the risks they are running. But no sooner had it been introduced than the financial crisis raised a fundamental question about this approach: how to measure those risks?
In March 2016, the Basel Committee on Banking Supervision took a significant step in this ongoing debate. It proposed removing internal models from the capital framework for operational risk – risks such as fraud, cyber failure or misconduct fines. Operational risk is only a small part of banks’ total exposures, typically 10% to 15% of total risk-weighted assets, compared with about 80% or more for credit risk. But the decision to remove internal models is a striking reflection of the direction of travel.
Is this the right way to stop regulatory arbitrage – the danger of banks running their internal models specifically to minimise capital requirements? Possibly, but removing internal models is no panacea. The alternative is a standardised formula for calculating risk. If the regulator makes a mistake with that standardised formula, then the entire banking sector is at risk, instead of just one bank with a faulty internal model. And it is very hard to devise a single formula that adequately reflects the starkly different banking environments in, say, Brazil and Sweden, which are both Basel Committee members.
There is a school of thought that says we have overcomplicated matters. The Bank of England’s Andy Haldane or Thomas Hoenig of the US Federal Deposit Insurance Corporation argue we should simply impose a very high core equity capital ratio on systemic banks, and then regulators can dispense with trying to micromanage how these banks measure risk. But as Mr Haldane’s colleague, Alex Brazier, recently pointed out, it is economically inefficient for banks to hold large surplus capital buffers if they don’t need to.
Mr Brazier’s answer is the counter-cyclical capital buffer: if systemic risk is rising, so should capital requirements. The tool has yet to be tested through a credit cycle, but it has one obvious merit. Supervisors can change such buffers overnight, whereas recalibrating regulations takes years. But there is also a downside supervisors will need to overcome. As Justin Pugsley writes this month (see page 66), investors urgently need more transparency on how supervisors calculate these capital add-ons, to avoid market shocks like the one seen in European coco bonds earlier this year.
