Bank capital regulations are notorious for introducing weird, distortionary effects. With successive versions of the Basel Accords, these have just got more numerous and baffling.
The securitization industry, for example, howled with anguish at insurance rules that make it more expensive for an investor to hold the safest 80% of a book of mortgages than to hold the whole thing.
Banks from some countries cannot compete in certain kinds of underwriting, or can no longer write swaps for certain kinds of client.
Start-up banks lending to the safest customers needed to hold more capital against their loans than established firms need to lend to self-certified deadbeats. Placing bank equity in an SPV cuts the cost of holding it in half.
But by far the weirdest effects come from operational risk rules, which actively damage the economy and undermine good behaviour among banks.
Like so many dumb ideas, they started with good intentions.
Alongside credit risk and market risk (the risk of lending going bad, or securities dropping in price) sits operational risk, the danger of human error. Not so much the “we lost money on that loan, oh well, that happens sometimes in underwriting” kind of error - more like rogue traders, fat fingers, fines, hacking, customer settlements and so on.
In “normal” times, capital to safeguard against such disasters accounts for about 10% of a universal bank’s balance sheet. But that was before the banks started getting fines in the billions.
Because it’s hard to measure the likelihood of business fumbles, idiocies or failures (credit risk and market risk can at least have quantitative inputs) the formula for operational risk is rigid and inflexible.
There’s idiosyncratic risk – what’s expected from your own business – historical risk, and industry-wide risk. It’s the last two, however, that give rise to perverse incentives.
Start with historical risk. This hangs around for five years after you’ve left a business behind. Since the banks that restructured most quickly, UBS for example, started doing this in 2011 and 2012, they should now be starting to roll off some of their operating risk capital.
But the incentive it creates is to keep crummy legacy businesses with op risk problems around for longer, rather than disposing of them. Selling or closing legacy businesses might free up capital associated with credit risk or market risk, but it also usually means losing revenue, or at least, closing off an option to benefit from a recovering market.
Restructuring an investment bank is a trade-off between the two imperatives – but being forced to keep the operational risk for years after closing the activity leans on the side of keeping businesses open when they are past their sell-by dates.
The industry-wide component also pushes in exactly the wrong direction. Investors, regulators, management and the public all want speedy, comprehensive bank settlements – solve the problems of the past and move on.
But industry-wide operational risk rules – essentially, hold more capital because your peer group has messed up – cut the incentives to settle fast, disclose, move on. They even cut the incentives for banks to help regulators investigate their peers.
The logic that a bank might run similar operational risks to its peers is not wrong in the abstract. But it is very wrong in the specifics.
If bank A choses to settle, say, all its residential mortgage-backed securities cases up front, act the model citizen, and hand over every piece of information related to RMBS mis-selling it can find, it will usually be rewarded with lower fines. But its operational risk costs will keep growing for years afterwards, as its peers stumble in with their own, larger settlements. It doesn’t matter that the odds of being fined again for the same offence by the same regulator are tiny – if industry peers get fined, operational risk gets expensive.
Barclays’ decision to fight the Department of Justice about US RMBS mis-selling, meanwhile, seems fraught with danger – not just for Barclays, but for its peer group, even banks that got out of US RMBS four years ago. A loss for Barclays, with a punitive settlement, will hurt Royal Bank of Scotland through operational risk, though it closed US RMBS underwriting in 2015.
The sums involved are not trivial. UBS reports Sfr223bn of risk-weighted assets at the end of December, of which Sfr77.8bn were from operational risk – meaning Sfr11.2bn of capital.
Forcing banks to tie up capital in op risk does make them less likely to fail – just as writing in an arbitrary percentage buffer for every bank would.
But if that’s the goal, pushing more capital on to credit risk or market risk would work better. The rules for those are more complicated, but they incentivise roughly the right behaviour.
Banks get capital benefits from showing they have good risk management systems, a good understanding of their assets, or hedges laying off some of their risk. Managers can pass these constraints on to businesses, encouraging desk heads and staff to take responsibility for how they manage capital.
Instead of scrapping op risk capital, though, the Basel Committee intends to replace it in something like the current form. There will be even fewer discretions and opportunities for modelling, and the calculation will be even more mechanical. Any easing will take the form of tweaking the number of years it takes for op risk events to fall out of the dataset, not ending the absurdities of the measure entirely.
Adding more capital to banks might be good or bad, depending on how it’s done. But op risk is clearly bad. The Committee should be bold and scrap it, not replace it.