Today’s New York Times editorials include a topic one might have thought beneath its august gaze: replacements for credit-rating agency references in a complex proposal rewriting the U.S. capital rules for market risk. The Times is right – the rule is in fact an important proposal to prevent a repeat of all the systemic risk wrought by over-reliance on indolent raters. But, the editorial is less well-founded when it fires off another blast at big banks, arguing they oppose the proposal just because banks liked it better when they could issue junky securitizations with nary a care. We won’t deny that there’s a bit of nostalgia on the Street for the happy days of frisky risk arbitrage. But, that doesn’t mean regulators got it right in this complex rule. In fact, one part of the proposal – the way risk would be judged for sovereign issuers – is so fraught with complexity risk that it could well sow more systemic risk than it seeks to cure.
The question of how to rate sovereigns is vital. We’ve learned the very, very hard way that the debt obligations of even the largest nations aren’t risk-free even though global capital and liquidity rules like to think they are. The zero risk weighting for sovereigns and the expectation that markets are liquid for them without limit led big banks around the world – not to mention most other financial institutions – to house many multiples of tangible equity into this sector. And, the contagion risk gripping the EU is only one ill effect of this risk-free treatment – the zero weightings also create an irresistible regulatory incentive for banks to hold big books of national debt instead of supporting credit demand in the private sector.
But, after agreement that sovereign debt isn’t always risk-free comes the hard task of differentiating it. Regulators could do this by just issuing charts setting specific capital standards country by country, but this would be so rude. Naming names might be taken amiss in another sovereign. So, the U.S. proposal tries another tack at objective sovereign weights without ratings. The inter-agency proposal turns to the “country risk classifications” (CRCs) developed by the OECD. On paper, the OECD’s system looks appealing – it’s apparently objective and seemingly rigorous. But, on closer examination, the CRC is an even less robust way to look at sovereign risk than the rating agencies.
Why? Quite simply, the CRC isn’t designed to judge credit risk. Instead, its goal is to set export-credit risk premiums based on currency convertibility risk, which has just about zip to do with whether a nation issuing the currency of the rated obligation can in fact make good on it. Under the CRC, Greece and Portugal are so “low risk” that they get the coveted zero risk weight. We know that the NPR seeks to correct for this glaring flaw with another provision, which tacks on a higher capital charge if a country has defaulted or otherwise skipped its payments. But, this is of course a retroactive risk weight that flies in the face of the whole point of these capital charges.
Regulators have made capital the bulwark of their reformed regime because it’s supposed to be a forward-looking buffer between bank risk-taking and ruinous demand on the taxpayer. Capital is to anticipate both the probability of default (PD) and the loss given default (LGD) on credit exposures not so expected that PD and LGD have been realized in addition to the bank’s loan loss reserves. In essence, capital is the price of unexpected risk based on what are hoped to be reasonable predictors of what is yet to happen. If capital only reflects realized risk, it’s as flawed as ratings that only recognized increased PD when a creditor hangs the “out of business” sign on the door.
Are there better ways to judge sovereign risk? Of course – that’s why not everyone has lost everything in the EU. The problem with many of these risk methodologies is, though, that they are intensely proprietary – that’s why people bet on them – and, often rely so much on market indicators that, if more widely adopted, market incentives would be distorted or subject to manipulation. The real solution to sovereign ratings is comparable to other cures to complexity risk: setting clear criteria for how much risk a bank may take for each risk weighting, disclosure of benchmarks for risk weightings and – most important – action by regulators when bank risk weights don’t pass the laugh test. Piling a new system atop old supervisory practice will only make it still harder to get banks back to the business of lending to consumers and businesses