Economics and similar, for the sleep-deprived
A subtle change has been made to the comments links, so they no longer pop up. Does this in any way help with the problem about comments not appearing on permalinked posts, readers?
Update: seemingly not
Update: Oh yeah!
Wednesday, April 10, 2013
Upside down CDOs and leverage ratios
I thought I might as well expand on the tweet quoted here, as rather too many people seemed to have jumped on the bandwagon of assuming that anything "risk weighted" is in some way suspect and probably manipulated, while "straight" leverage is honest, simple, probably speaks with a Northern accent, et cetera.
Not so many people realize that under the Basel standards, you can often get risk weights of more than 100% assessed, even using those terrible internal models. This is because (as I alluded in that tweet), the question of what appears on a balance sheet is not a fact of nature - it's a control parameter itself.
Stretch your memory back to 2007(!) and Felix's explanation of how to make a CDO if you need to. Then consider a really simple transaction:
1. I'm a bank and I take 1000 loans, each with an expected loss of 5%, so the expected loss is 50.
2. I bundle them up and sell them to a special purpose investment company.
3. That special purpose investment company has two classes of debt - 500 of senior and 500 of subordinated.
4. I sell the 500 of senior to bondholders and keep the 500 of subordinated.
For reasonably normal accounting policies, my balance sheet will now show that I have sold 500 of my 1000 of loans, and will be smaller by 500. An accounting balance sheet is always going to show something like this, because the 500 that I've sold off are now on someone else's balance sheet and so shouldn't be on mine.
The risk-weighted balance sheet, though, will show no effect at all.
The regulator would argue thus:
1. The expected loss on this portfolio is 50.
2. The senior debt does not take any loss unless the realized loss on the loans is 500, ten times the expectation.
3. Therefore, the junior debt takes all the realized loss in all but a tiny minority of situations
4. Therefore, the 500 that is on my balance sheet is bearing substantially all the risk of the original 1000 of loans.
5. So the 500 asset on my balance sheet needs to be given a risk weighting of 200%.
Of course, as the man said "It's More Complicated Than That", but this is the idea. A regulation based on accounting leverage is very easy to game because the composition of the balance sheet matters. By the way, if you think that this is all a product of nasty investment banking and financial wizardry and could be stopped by sufficiently draconian Glass-Steagall type law, no luck there either.
All that is really going on here is that the leverage ratio doesn't distinguish between a loan to an ungeared borrower and a loan to a borrower that has loads of other debt. So it doesn't have to be a special purpose vehicle although that makes the problem clearer; historically it has been found that leverage regulations reliably give you a banking system with no surplus cash balances and loads of leveraged buyout loans. One of the things that they teach you in the relevant business school course is that you need to dig down to the leverage of the underlying "real" assets, and that's basically what Basel risk-weighting is all about.
There aren't any silver bullets. There's no substitute for an active, involved and self-confident supervisor. But "straight" balance sheet leverage is a real blind alley. It's a classic case of taking a set of information produced for one purpose and hoping it will be useful for another. As far as I can tell, the idea that people have is that they want "objective" measures, based on the financial accounts because they don't trust the banks not to fiddle the model. But if that's your problem, then that's your problem right there - if they're meant to work as a solution to a fundamental problem of trust, then leverage ratios are ludicrously inadequate.
 Pet theory alert; you'd be surprised at the extent to which a lot of financial regulation is all about the sublimated performance of masculinity.
 There are echoes here of the 2008-era religious horror of "Level 3" or "Mark to Model" assets. As in that case, refusing to model something (or to accept someone else's modeling of it) doesn't actually make the answer more objective, it just ensures that you will remain ignorant whatever information arrives.
 I have no idea how anyone says this with a straight face, and suspect that the answer might be lack of ever having seen a balance sheet prepared.
this item posted by the management 4/10/2013 12:41:00 AM