Over the weekend, the Basel Committee will continue work on Basel III. This suggests to us a cautionary tale of why the best-laid schemes of regulators don’t always work out so hot.

The scheme in question is the operational risk-based capital (ORBC) part of Basel II. We worked hard to keep this out of Basel II on behalf of several clients and, we believe, a general interest in making regulatory capital make sense. We argued at the time that operational risk had, at best, questionable measurement methodologies and quantification techniques, meaning that a stiff regulatory charge for an admittedly material risk was premature, if not misguided.

But, we lost and Basel II includes two provisions for operational risk. One is, we think, flat-out silly: a capital charge based on a simple percentage of assets prone to operational risk (e.g., those under management) that penalizes banks no matter how good their operational risk-management or mitigation might be. This is called the “standardized” ORBC charge. The second option is the advanced measurement approach (AMA), a more nuanced approach based on internal models.

Consistent with its decision only to allow U.S. banks to use the fanciest parts of Basel II, the U.S. rules include only the AMA. Readying the U.S. for the rules – years late, of course, but everyone’s been busy – the agencies actually issued new guidance on the AMA a few weeks ago.

Which brings us back to the cautionary tale. Operational Risk & Regulation earlier this week reported that Lloyds Bank in the U.K. abandoned the AMA after several years of trying as hard as it could to implement the cursed thing. The problem in part derived from the bank’s acquisition of another troubled U.K. lender before things turned far worse and the combined firm was nationalized during the crisis.

ORBC isn’t easy – that’s why we argued against it because of all the underlying questions about how to measure and map it. So, it takes years, not to mention lots of dollars, to craft an ORBC model that at least meets the laugh test. When Lloyds acquired HBOS, it got that bank’s model and, then, the task of trying to combine it with its own. Years later, the bank gave it up and, now, has gone back to the standardized approach.

But, the standardized approach actually doesn’t make a lot of sense – the reason the U.S. rightly decided against it. However, the standardized ORBC charge gave Lloyds a number. The AMA gave it nothing but grief.

We know that some U.K. banks have instituted some version of the AMA, but the vast majority of banks under Basel II begged to differ. They went for the simple, albeit, if irrelevant, standardized charge. And, those banks that have bought into the AMA have largely done so with considerable waivers from their regulator designed to ease the passage to the AMA without doing much to ensure that the capital charge in fact makes any sense at all.

Can the U.S. make better use of the AMA now that regulators have told us how to proceed? That, of course, remains to be seen, but the total reliance on internal models combined with the acknowledged inability of anyone to calculate the meaningful equivalent of a leverage charge for operational risk makes this, at best, iffy.

We aren’t questioning the importance of operational risk – as recent events have shown, it can threaten even the seemingly soundest of banks. We are, though, arguing against uncritical reliance on untested models on which costly capital requirements are based that, in turn, pile on to still other capital charges into a heap of costly standards that will combine in, as yet, unknown ways at a most delicate time in global financial markets.