Were You To Do So

Every working day for about thirty years, a small number of bankers in London were asked the same question, and their answers set the interest rate on a very large share of the loans on earth.

Here is the question. At what rate could you borrow funds, were you to do so by asking for and then accepting inter-bank offers in a reasonable market size just prior to 11 am?

Read the middle of it again. Were you to do so. The question does not ask what the bank borrowed at. It asks what it thinks it could have borrowed at, in a transaction that need never have taken place, in a size nobody defines.

What rested on the answer

The London Interbank Offered Rate was the number other numbers were built on. Corporate loans, mortgages, student debt, and above all interest rate swaps, where one side pays fixed and the other pays LIBOR plus a margin.

When the British government finally reviewed it, the report put the figure plainly: contracts with an outstanding value of at least 300 trillion dollars reference the benchmark. That is several times the annual output of every economy on earth, hanging off a set of daily estimates.

How the number was made

Each panel bank had an application on a desk in its treasury. Between eleven and ten past eleven every morning, one person at each bank typed in that day's rates and sent them to Thomson Reuters, the calculation agent. No bank could see another's submission during the window. The rates were published at noon.

The submissions were sorted, the top quarter and the bottom quarter thrown away, and the remaining half averaged. On the main dollar panel of eighteen banks that meant discarding the four highest and the four lowest and averaging the ten in the middle. That is a trimmed mean: it stops any one outlier, honest or otherwise, from dragging the result.

It was done for ten currencies and fifteen maturities, from overnight to twelve months. One hundred and fifty rates a day.

Moving the number therefore requires moving several submitters at once, or moving one enough to change which submissions get trimmed. Both turn out to be possible if the submitters talk to each other, which is what the enforcement cases from 2012 onwards established that some of them did.

Manipulation is the part everybody already knows, and it is not the most alarming thing on the record.

The BBA said so itself

The most damaging sentence about LIBOR was published by the organisation that produced it, on its own website, explaining the rate to the public.

bbalibor is not necessarily based on actual transactions, it read, as not all banks will require funds in marketable size each day in each of the currencies and maturities they quote, and so it would not be feasible to create a full suite of LIBOR rates if this was a requirement.

That is not a leak or an allegation. It is the administrator explaining, reasonably and in plain English, that a hundred and fifty rates a day cannot all be backed by trades because there are not a hundred and fifty trades a day to back them. The page went on to defend the practice: a bank knows its own credit and liquidity risk profile from the rates at which it has dealt, and can construct a curve to predict the correct rate for currencies or maturities in which it has not been active.

The same page conceded that reasonable market size, the phrase in the middle of the question, is intentionally left broadly defined.

The wording changed in 1998

It had not always been a question about yourself.

Until a review in 1998, panel members were asked something different: at what rate do you think interbank term deposits will be offered by one prime bank to another prime bank for a reasonable market size today at 11am. That is a question about the market. It asks a banker to observe what is going on around him and report it, and if he is wrong, tomorrow's dealing will show it.

The replacement asks what he himself could get. The BBA's stated reason for the change was that the new definition enables accountability for the rates, and the logic is not silly: tie each submitter to his own institution's cost of funds and he owns his answer rather than gesturing at an average.

The side effect was that a bank's daily submission became a public statement about what the market thought of that bank. During 2008, when a high submission would have been read as a confession that nobody would lend to you cheaply, that was a powerful reason to send in a number that was not strictly an answer to the question.

Nobody was actually in charge

Here is the sentence from the Wheatley Review that should have caused more of a fuss than it did.

At present neither submitting to LIBOR, nor administering LIBOR, is a regulated activity under FSMA.

The Financial Services and Markets Act 2000 is the statute that decides what counts as a regulated financial activity in Britain. Advising somebody on a pension is regulated. Arranging a mortgage is regulated. Producing the interest rate that sat underneath three hundred trillion dollars of contracts was not, and neither was submitting to it.

The regulator was pursuing manipulation cases anyway, but it was doing so sideways, through the general conduct rules that applied to the banks for their other business. There was no rulebook for the thing itself. The rate was compiled by a trade association, the British Bankers' Association, on behalf of its own members, who were also the ones supplying the inputs and the ones whose contracts were priced off the output.

The market it was meant to measure had thinned out

LIBOR was meant to summarise a real market: banks lending each other unsecured cash for a month, three months, six months. After 2008 that market shrank, because the lesson of the crisis was that you might not get it back.

The Review is careful but clear. The market for inter-bank deposits, it says, is not transparent, and LIBOR users are not necessarily aware of the volumes of the inter-bank transactions that underpin the benchmark. It printed a chart from the European Central Bank's money market survey showing unsecured euro lending and borrowing falling from a peak in 2007 to roughly half that by 2010.

So the submitter faced a question about a transaction that not only did not happen, but for that currency and maturity might barely happen at all any more. The estimate was always an estimate, and the thing it estimated was draining away underneath it.

What the review did about it

Martin Wheatley reported on 28 September 2012 with a ten point plan, and its logic runs one way throughout: turn a question into a measurement.

Submitting and administering should become regulated activities. The BBA should hand the rate to an independent administrator. Banks should use transaction data to corroborate what they send in. Individual submissions should be published after three months, so a bank pitching a number that suits its book has to face it again later.

And the one that matters most: stop publishing the rates that cannot be supported. Compilation should cease for those currencies and maturities with insufficient trade data behind them. Not fix them. Stop producing them.

It declined to abolish LIBOR, and said why. Ripping it out would have meant an unacceptably high risk of financial instability and large scale litigation between parties holding contracts that referenced it. The plumbing was faulty and the building was already on top of it.

The slow ending

The rate went in the end, by timetable rather than collapse. The sterling, yen, Swiss franc and euro panels stopped at the end of December 2021, and around that the clearing houses moved more than 13 trillion pounds of sterling contracts to SONIA. The remaining dollar panels ceased at the end of June 2023.

What replaced it corrects the original design. SONIA and its American equivalent are built from overnight transactions that actually occurred, compounded over the period you need. Nobody is asked what they could have borrowed at. The rate is a record of what was paid.

What the question was really for

Be fair to whoever wrote it. In 1986 the interbank market was busy and a treasurer knew what he could get funds at that morning, because he had done it the day before. The question was a shortcut to a fact, and a shortcut to a fact does not announce the day it stops being one. The wording did not change after 1998. The market underneath it did, slowly and then very fast, and the question went on being asked at eleven every morning in the same words, of people with less and less to answer it with.

Three hundred trillion dollars of contracts were priced off the conditional tense.

Sources

• HM Treasury, "The Wheatley Review of LIBOR: final report", 28 September 2012 (the estimate that contracts worth at least 300 trillion dollars reference the benchmark; the finding at paragraph 2.4 that neither submitting to nor administering LIBOR is a regulated activity under the Financial Services and Markets Act; the ten point plan, including statutory regulation of both, transfer of the rate from the British Bankers' Association to a new administrator, use of transaction data to corroborate submissions, publication of individual submissions after three months, and cessation for currencies and tenors with insufficient trade data; the conclusion that reform was preferable to replacement because a transition would pose an unacceptably high risk of instability and large scale litigation; paragraph 5.35 on the inter-bank deposit market not being transparent and users not knowing the volumes underpinning the benchmark; and Chart 5.A, from the European Central Bank Euro Money Market Survey 2011, showing unsecured euro turnover falling from its 2007 peak to roughly half by 2010).
• British Bankers' Association, "bbalibor explained: the basics", text of the BBA's own bbalibor.com pages (the submission question in full; production for ten currencies and fifteen maturities from overnight to twelve months, giving 150 rates each business day; the statement that bbalibor is not necessarily based on actual transactions because not all banks require funds in marketable size each day in each currency and maturity they quote, and that a bank can construct a curve to predict the rate where it has not been active; the concession that reasonable market size is intentionally left broadly defined; the pre-1998 wording about what one prime bank would offer another, and the statement that the new definition enables accountability for the rates; the 11:00 to 11:10 submission window, Thomson Reuters as calculation agent, publication at noon, and submitters being unable to see each other's rates; and the trimming table by panel size).
• John Kiff, "Back to Basics: What Is LIBOR?", Finance and Development, International Monetary Fund, December 2012 (the eighteen banks on the main US dollar panel and panels of between six and sixteen banks for other currencies; the ten currencies; the fifteen maturities from overnight to one year; and the discarding of the four highest and four lowest submissions before averaging).
• Financial Conduct Authority, "Announcements on the end of LIBOR" (the cessation of the sterling, Japanese yen, Swiss franc and euro panels at the end of 2021, and the arrangements for the wind down of the benchmark).
• Financial Conduct Authority, "The USD LIBOR panel ceases at end-June 2023" (the cessation of the remaining US dollar panels at the end of June 2023).
• Financial Conduct Authority, "Finalising LIBOR transition: achievements in sterling markets and what remains to be done" (the conversion of over 13 trillion pounds of LIBOR referencing contracts to SONIA through central counterparty conversion in December 2021, and the embedding of overnight SONIA compounded in arrears across sterling markets).