Main index Financial Markets index About author

Libor, buttered side down

Julian D. A. Wiseman

Abstract: the US authorities want Libor to be fixed using alternative rates “anchored in observable transactions”. But, despite the authorities’ current interest in a transactions-derived Libor, it wouldn’t fix the problems, and might make them worse.


Publication history: only at Usual disclaimer and copyright terms apply.

Contents: Introduction; As smooth as buttered side down; Payments; Conclusion; Footnotes; Afterword.


US regulators are proposing that Libor be fixed using alternative rates “anchored in observable transactions” (FT, US regulators urge quick Libor replacement, 25th April 2013). While the idea is superficially attractive, it just won’t fix the core problem. The problem is that the market in term unsecured inter-bank loans is so thin that it cannot take the burden of fixing the huge quantity of outstanding Libor derivatives. Indeed, switching to a transactions-based Libor would hugely increase the number of entities able to manipulate the fixings. To justify this we need to discuss the the relative size of the markets in the underlying and the derivatives.

 As smooth as buttered side down

A farmer and a banker wish to trade the price of butter. As the farmer sells butter to UK supermarkets, the farmer and banker agree to trade the price of a particular butter in a UK supermarket. The farm is to sell, the banker is to buy.

A price is agreed, and a ‘fixing date’, that is, the date on which the price of butter is to be observed. Also a size: and as this is only a practice trade, to test the systems, the size is only 100 packs of butter. So for each penny the price is below or above the trade price, the seller or the buyer receives £1 from the other.

And on the fixing date they go to the agreed supermarket, observe the price of butter, and settle. So far, so good.

But that was in a hundred packs. What if the trade was in size of millions of packs, or even hundreds of millions. The bank would not want to carry that risk for a long time, so after trading with the farmer would endeavour to find a buyer of butter, to whom to on-sell the position. Assume that this second trade has a later fixing date.

So in the morning of the fixing date of the trade with the farmer, the banker is about flat, neither long nor short. But once the settlement price for the trade with the farmer is fixed, the banker will be short. Very short. With that imminent, the banker should be buying butter. But the banker buying butter might raise the price of butter. The farmer was short, and would lose from this increase in price. Was the bank fiddling the settlement, or was the bank fairly hedging?

Sometimes it is easy to tell, most easily when ill intent is admitted in an email or a phone recording. (Yes, people can be that stupid.) But based on just the behaviour, the honest hedge and the dishonest fiddle involve much the same action.

The problem here is that hedging the fix will move the market in the underlying. This is because the derivative position has been assumed to be much larger than typical trading in the underlying. This size mismatch is also true for transactions against Libor.

So why is the derivative market traded in sizes so much larger than can be absorbed by the market in the underlying? It is easy to assume that it is all the fault of evil speculators. Easy to assume, and morally reassuring, but wrong, as can be seen from a glimpse at the market in inter-bank loans.


Widgets Plc, which banks at HSBC, has sold widgets to Systems Plc, which banks at Barclays. Systems Plc must pay £1bn to Widgets Plc. So Systems Plc instructs its bank to pay. Barclays lowers Systems Plc’s account by £1bn, and HSBC adds £1bn to the account of Widgets Plc. All that remains is for Barclays to pay HSBC. Both banks have a £ account at the Bank of England, and Barclays instructs the BoE to transfer £1bn to HSBC. The payment is done. (Of course, a real payment would be in a slightly different order — irrelevant for this story.)

So Barclays now has £1bn less in its account at the BoE than it did a few moments previously, and HSBC £1bn more, and this was not caused by a decision of either bank. Their customers compelled it. And if both banks previously had about the right amount of money to hand, one now has too little and the other has too much. The obvious remedy is for the other to lend £1bn to the one. In some sense, the market in inter-bank loans exists to offset customer payments.

But there is more than one payment each day. CHAPS, the UK’s large-value payment system, has a daily average of ≈135k payments with total value of ≈£285bn. Much of this value will cancel out: Barclays pays HSBC, HSBC pays RBS, RBS pays Lloyds TSB, Lloyds TSB pays Barclays, all for different customers. A bank’s mismatch is the difference between its receipts and payments.

So if many of these large non-banks had hedged their interest-rate risk in the derivative market, but the total of client payments was nearly in balance for most banks on most days, then the derivative market could be much larger than the inter-bank loan market, even without speculators. And the speculators, who typically make markets more efficient, are almost entirely in the derivative market.


So the ideal is that the market in the underlying be large relative to that in the derivative markets. ‘Large’ meaning that, as a derivative is fixed, one could trade in about the size of the fixing, and not move the market in the underlying. Rephrased, the ideal is that the market in the underlying can take the weight put upon it.

In the mid-1980s, when the BBA started its Libor fixings, the inter-bank loan market was large and active, and the derivatives were few and small. So it worked, at least mostly. But post credit-crisis, there is a huge volume of derivatives outstanding, while the inter-bank loan market is smaller, and in some maturities almost non-existent. The market in the underlying is just too small relative to the weight put upon it.

Observe that this whole argument does not distinguish between a Libor based on estimates, and a Libor based on transactions. Relying on transactions, few and small as they are, might or might not be better, but will definitely be inadequate. (Indeed, there are reasons to believe that it would be worse, but that isn’t important to this argument.)

This essay might seem to be a council of despair: nothing will work. Not quite: there are more radical solutions that would be much more robust. But a transactions-derived Libor will not fix the problem, even if it appears to change it.

— Julian D. A. Wiseman
London, 1st May 2013


The Bank of England’s Payment Systems Oversight Report 2012, page 18.

Currently banks estimate the price at which they could borrow. Problems with those estimates have been reported, but now, with fines and prison sentences in clear view, all participants in the process are being paranoid. There is the problem of measuring something that barely exists, but it is honestly done.

Next, assume Libor becomes transactions-based. So a hedge fund could quietly accumulate a cash pile, and just before the fixing of a massive derivative position, start lending the money cheaply. Very cheaply. Transactions would be happening, in the relevant currency at the relevant maturity, at this depressed rate. And so that is where a transactions-based Libor would fix. The (‘received’) derivative position would win, and if big enough, would win far more than the losses from lending cheaply. And of course, this could be done by a central bank or other entity with the privilege of sovereign immunity (some of whom used to be aggressive participants in the FX market near knock-in or knock-out triggers). This would not be much of an improvement in the functioning of Libor.


This and other comments were discussed on the FT on 1st May 2013: Dumb and dumber.

Main index Top About author