I assume that there's someone somewhere whose job it is to think about this, but the big Libor fine that appears to be in UBS's future got me wondering: how do they decide how big these fines are supposed to be? In most fraud cases you can tot up how much someone stole and use that as a starting point, inflating or deflating it for different levels of evil or remorse. But that doesn't seem to be a promising avenue in Liborgate, where the money involved is hard to calculate and mostly flowed around the manipulating banks without touching them directly. The fine-setters seem to have about four things to think about:
- how much bad stuff did the bank do,
- how much money did they make doing it,
- how caught are they, and
- how sorry are they now.
On how much bad stuff … really the point of these settlements is that you'll never quite know. The Barclays settlement documents contain tons of delightful emails, but they're framed by the usual prosecutorial boasting that they are "just some examples of the numerous trader requests over the years in question." They're a sampling thrown in for scandalous effect, not a real accounting of Barclays' rate manipulation. For the CFTC to actually publish every instance that it discovered of rate-fixing, in a settlement, would be silly. For one thing, the settlement is designed to avoid the necessity of doing the work to get such an accounting. For another, the settlement is designed to avoid the public release of such an accounting, which would be ammunition for the private lawsuits that have sprung up around Libor.
So we're unlikely to get a real official read on whether UBS was worse than Barclays and by how much. But the fine is obviously a clue that they were pretty bad. From David Merkel's data they actually seem to have been middle-of-the-pack as a Libor submitter, without the extreme submissions and big swings that Barclays had. But to be fair that is in 3-month USD and part of UBS's thing seems to have been manipulating Libor in more tenors and currencies than Barclays did.
On how much money they made … you'll definitely never know that. That hardly even seems like a thing that is knowable, but of course banks do measure their interest rate sensitivities and you can measure the effect of a screwy Libor submission so, yeah, I guess UBS could figure it out if they wanted to try really hard. (They don't, of course; thus the settlement.) I did one simple-minded exercise and tried to figure out how much interest rate exposure each bank had, based on reported VaR numbers; here is what I got:1
So, loosely speaking, UBS had on average about as much exposure to interest rates as Barclays did, or maybe a little less, over the big Libor-manipulating times. Which suggests that, if Barclays and UBS did about the same amount of manipulating with about the same success, then they'd have made about the same amount of money doing it. A fine of two Barclays or so, then, suggests that UBS did at least twice as much manipulating.
On how caught … these settlements are a negotiation in the shadow of "just sue us," so the more evidence the regulators have against you the more money they can extract, even though that doesn't have much to do with culpability. As Felix Salmon said, "Barclays is being hammered here, and its CEO has lost his job, not because Barclays was worse than anybody else, but rather because its employees were stupid enough to leave a written record, in emails, of exactly what they were doing." I'm sure I speak for all of us in saying that I hope UBS has Libor emails that are as juicy as Barclays', but that's not strictly necessary for the regulators: UBS basically turned itself in.
Which gets to how sorry they are. You can read about the litany of immunities and leniencies that UBS has gotten from various regulators in exchange for its voluntary cooperation. That, and the fact that it's the second settlement, might justify a fine somewhat lower than Barclays'; the fact that the fine seems to be much larger again suggests that UBS was a much worse manipulator. Which I guess is good news for the prospect of silly emails.
Anyway, feel free to use this fake model to think about the next shoe to drop. Everyone thinks that's RBS; from this Bloomberg article they seem to have been pretty bad, and in a blatant-email kind of way, though they had less money at stake on interest rates than either Barclays or UBS did. The fact that they're owned by the British government also probably helps; it's hard for governments to fine themselves, or their other-government friends, and it suggests that any Libor manipulating done by RBS can't have made them all that much money.
1.This was a fun little thing to put together. Barclays reports daily interest rate VaR at a 95% level in GBP; UBS reports 10-day "Interest rates (including credit spreads)" VaR at a 99% level in CHF. So UBS reported 397mm CHF of rates-'n'-spread VaR for 2008, for instance, which I simmered down to $34.7mm of 95% daily rates VaR through a series of dodgy operations set out here. Dodgiest is the split between rates and credit, which I just did by looking at that split at Barclays and RBS and assuming UBS was in the middle of them.