Measures, good and bad July 2, 2013 at 8:25 pm
The story of Thomas Hoenig’s `horrible’ assault on Deutsche got me thinking about multiple risk measures. Here’s the thing:
Long, long ago, we believed, or at least set up the regulatory framework as if we believed, that we could design a single measure of how safe a bank was. It was the RWA calculation.
More recently we have admitted that we can’t do that, and instead demand that banks meet a number of criteria: capital, leverage, LCR, and so on.
There are two problems. One is knowing when to stop. At some point, arguably already, the framework gets so complicated that no one knows what it does. There is an argument that less is more.
The second problem is that while having more ratios can prevent a false negative problem, as they backstop each other to some extent, they don’t prevent a false positive problem. The new leverage ratio, for instance, to put it charitably suffers from some design flaws which cause it to dramatically over-state risk*.
The key I think is that the backstops like the leverage ratio† need to be genuine backstops, and not new constraints for most firms. Then attention can be focussed on the one or two key measures which should work for most firms, and all large ones.
*This might partially explain Hoenig’s remarks.
†The leverage ratio as originally conceived was a backstop. Now you might argue – and Hoenig certainly would agree – that it should play a central role and be set much higher. That’s fine, but in that case you need something to backstop the cases where there aren’t many assets but they are all really risky.