What’s wrong with models? Nothing regulators can’t fix if they can fix themselves. The heart of Basel’s jihad against the internal ratings-based (IRB) approach in the Basel risk-based capital rules lies in the new-found belief that models are the primrose path to variability, complexity, and regulatory arbitrage. However, variability is necessary because nations and assets aren’t all the same, complexity is required to capture forward-looking risk, and regulatory arbitrage is only possible if regulators let banks play the numbers. If regulators know they can’t trust themselves or their colleagues, they can as they have proposed make models anathema. However, doing so only makes banks riskier.
Earlier today, we sent you an in-depth analysis of Basel’s latest effort to eschew models in the credit risk-based capital rules. This follows our earlier assessments of similar efforts for market and operational risk. Each of these standards blazes the path to models-free capital, but also has its own provisions that, looked at in isolation, are simply astonishing. Stripped of the broad policy brush applied to the anti-models campaign, these seemingly-technical provisions show just how far afield Basel is going from anything that makes sense.
What are these “astonishing bits and pieces?” The complexity in Basel’s supposedly-simplifying standardizing proposals obscures them, but some of them are real jaw-droppers. Let’s start with the new credit-risk proposal, which bans the IRB for all but a few favored asset classes. Why are these allowed to touch the IRB and others consigned to the outhouse? Basel says it’s because the in-IRB assets could be arbitraged without models but the out ones are so low risk that banks can trust themselves to the determinations of credit rating agencies (CRAs) and the market.
Remind you of anything like all the lessons we said we learned after the crisis about CRAs and market risk pricing? It should. Essentially what Basel is saying now is that the banks they govern can’t be trusted to spot risk but the rating agencies that no one really regulates to this day are the font of forward-looking risk wisdom. And, if they aren’t, market pricing will guide us – a first, especially in these yield-chasing days. In short, banks are bad, regulators can’t make them better, and we must resort to a higher authority. That every bank must then bow down and rise at the same time doesn’t appear to bother Basel even though the wave this correlated risk could create would surely exacerbate systemic stress.
The operational risk-based capital approach to standardization is no better, as I detailed in a memo a few weeks ago. To recap, the Basel proposal sets standardized capital by two indicators – one based on income, which has nothing to do with risk, and another based on the past ten years of operational loss which has nothing to do with losses to come unless a bank is particularly accident-prone and its regulator is still more obtuse.
What about the market-risk capital standards now embodied by Basel in its fundamental review of the trading book (FRTB)? One might say that the FRTB bucks the anti-models mood because it’s based on a model called expected shortfall (ES). In essence, Basel has swapped a model it understands for the old one (value-at-risk or VaR) that it doesn’t like anymore. Presumably, because ES derives from the hand of the regulators, it is purer than VaR and it surely is simpler, at least in some respects. However, it’s never been tested and most banks think it doesn’t really capture their actual risk. Ironically, ES also has a lot of models embedded in it that will even be more challenging to banks and their regulators than VaR because no one has ever used ES and, as noted, it’s never been tested under fire.
Are models susceptible to subterfuge? Of course, but then so is everything else anyone does that’s got money involved with it. Far simpler things than models have gotten banks into way-big trouble. Simple mortgages precipitated the 1980s crisis and old-school commercial real-estate loans did their bit to blast banking in the early 1990s. Barclays blew itself up not due to complex risk models, but rather to simple, old-fashioned thievery. Long-Term Capital Management was a complex institution during the late 1990s, but the systemic risk that forced the Fed to engineer a bail-out wasn’t due to its holdings, but rather to big-bank risk concentrations in this high-flying fund. Up to the day of the financial crisis, most of even the biggest banks (at least the ones that then were banks) had limited exposure to the most complex ends of the subprime mortgage debacle, largely holding this soon-to-be-deadly risk in whole-loan positions or AAA tranches (darn those rating agencies!).
In short, simple doesn’t mean safe and complex can mean right. The way to tell the difference is not by imposing arbitrary fiats against complexity because the IRB makes the regulators’ heads hurt. It’s by making models better through backtesting, validation, and governance. To be sure, regulators tried this and banks complained about all the burden these requirements imposed – considerable indeed. Bankers failed to see that they had a choice – help regulators understand complexity or learn to live or die by simple edicts that essentially make no sense.
One more point about who loses when models are banned. Regulators are trying to save themselves by skewering models that they despair of supervising, but doing so exposes a fundamental flaw in the fragile framework of global regulatory requirements. The sum total of all of the changes to the risk-based capital rules shows that Basel has gone so far afield from its initial mission as to make itself not an arbiter of meaningful cross-border capital standards, but rather a committee increasingly reconciled to lowest-common-denominator consensus that lends credence to the concept of global capital standards instead of actually making them matter. Basel is used to demanding that banks take tough actions – time for Basel to do the same.