How to set Libor in a post-Barclays age

Time for the banks to have some skin in the game

Barclays. Rates up, up, up? Photograph: Getty Images

The Libor-fixing scandal is obviously problematic for a number of reasons, but one of the biggest ones is that it isn't entirely clear how to prevent it happening again. The only reason we know for sure the rates were artificially set are the (still astonishing) internal emails from Barclays requesting that.

The issue is that, certainly at the height of the financial crisis, and even now, Libor relies on self-reported rates. Banks rarely actually borrow in large quantities from the inter-bank market any more, so rather than reporting what they are borrowing at, they are forced to report what they think they could report at. The problem then is that unless you have cast-iron evidence that there is an ulterior motive, its hard to distinguish rate-fixing from just not really guesstimating your rates very well.

Some have responded to this by arguing that the replacement to Libor should be based on actual borrowing. If a bank isn't borrowing on the inter-bank market, they report some other similar cost of capital; if they aren't borrowing at all, they don't submit. But in the FT today, Frank Partnoy comes up with another, more ingenious solution:

The teeth of the new regulation would be a rule requiring the bank that submitted the lowest Libor estimate to lend a significant amount of money, say $1bn, to the Libor Trust at its submitted low rate. Conversely, the bank submitting the highest Libor estimate would be required to borrow the same amount from the Libor Trust, in the relevant currency for the specified period of time, at its submitted high rate.

So if Barclays under-reports its rate by 1 per cent and finds itself the lowest reporting bank, it suddenly loses out on $10m over the course of a year. And the same problem if it over-reports. If, meanwhile, RBS over-reports by 1 per cent, it loses out on the same; and the trust in charge of setting the rates pockets the $20m difference (which could then go towards running costs, into the Treasury, or some other noble cause).

If there is a problem with that, it's that it may not be entirely painless to be forced to borrow $1bn even at accurately reported rates. But that seems like a small price to pay for regenerating trust in a system which has suffered a severe blow.