Ideas for Fixing Libor

Many thanks to Stanford's Darrell Duffie for leaving a comment on my last post about Libor. Duffie is one of the academics who signed off on the WSJ's attempt to prove the measure flawed, so he's probably disappointed that the BBA has decided to do nothing, for the time being, to change it. I, on the other hand, think the BBA made exactly the right decision. Yes, there are problems with Libor, just as there are problems with any market yardstick. But the problems aren't huge, and rushing to fix them might well involve unintended consequences - not least that Libor rates, which are generally considered too low, might actually fall further.

Duffie talks of a few proposals for fixing Libor suggested by Terry Belton Bolton of JP Morgan. In fact, however, after running down the list of proposals, Belton concludes that none of them is a particularly good idea, on the grounds that they would generally either have negligible effect or that they would actually drive Libor lower. Still, it's an interesting list, and worth sharing. And given that the BBA has said that they "will be strengthening the oversight" of Libor, whatever that means, it's even possible that one or two of these might be adopted.

• Increase the size of the panel. At the moment, it is quite heavily UK-focused. But if more US banks were added, what would happen? The obvious ones, like Wachovia and Wells Fargo, tend to fund through deposits rather than in the wholesale markets. At the margin, that would tend to drive Libor down, since the banks in it would on average have less demand for interbank funds.
• Change the time of the quote. This seems to have been dreamed up by US bankers who don't like having to be alert at 11am London time. But, says Belton, "with a global index and a round planet time will always be a problem for somebody," especially since 13 of the 16 banks on the panel are based outside the US.
• Change the calculation to use a median rather than a trimmed mean. At the moment, the BBA discards the four most expensive and four least expensive quotes, and takes the average of the rest. Which is one reason why the attempted WSJ takedown was underwhelming: if the banks with the most credit risk reported higher borrowing costs, they'd likely be discarded from the calculation anyway. But what would happen if the BBA just used the median number instead? Belton did the math, and concluded that Libor would rise by less than one sixth of a basis point in crazy times like now, while "under more normal circumstances, such as in early 2007, we expect the statistical difference between the two measures would be statistically imperceptible."
• Change the survey question. At the moment, banks are asked what it costs them to borrow money on the interbank market; the Euribor survey, by contrast, simply asks what the interbank rate is. Would this have much of an effect? It's easy to answer that: just look at the difference between Euribor and euro Libor. It turns out to be about a quarter of a basis point: there's "no statistical significance to the different averages," says Belton.

It's doubtful that the first proposed change will happen: if it were going to happen, it would have been announced by the BBA on Friday. The others are basically tinkering at the margin, and would make Libor behave a bit more like Euribor. I suspect, however, that if Euribor had the amount of scrutiny of late that Libor has come under, it, too, would have shown weaknesses. Unless someone can compellingly demonstrate that Euribor is superior to Libor, both in theory and in practice, I see no reason to change Libor.