Libor Was Set by Asking Banks What They Would Pay
A benchmark referenced by hundreds of trillions of dollars in contracts was produced from estimates that banks submitted themselves. The design invited the manipulation that eventually surfaced.
How It Worked
The London Interbank Offered Rate was intended to represent the rate at which major banks could borrow from each other. It was produced each day by asking a panel of banks a question: at what rate could you borrow funds, were you to do so, in reasonable market size.
Submissions were collected, the highest and lowest were discarded, and the remainder averaged.
Note the wording carefully. Banks were not reporting transactions they had executed. They were estimating what they believed a hypothetical transaction would cost. It was a survey of opinion, and it became the reference rate for an enormous volume of contracts including mortgages, corporate loans, and interest rate derivatives.
The Two Distinct Manipulations
The first involved derivatives traders. Because trading positions were worth more or less depending on where Libor set, traders asked colleagues responsible for submissions to adjust them slightly. A small movement in the rate could be worth a great deal on a large derivatives book.
The second was reputational and occurred during the financial crisis. A bank submitting a high borrowing rate was signalling that other banks charged it more, which implied doubts about its creditworthiness. Several banks submitted rates lower than they believed accurate to appear healthier than they were.
The rate that priced hundreds of trillions in contracts was an estimate submitted by parties with positions that depended on the answer.
Why the Design Was Flawed
The structural problem was not merely that individuals behaved badly. It was that the benchmark asked interested parties for an opinion and had no mechanism for verification.
The problem worsened as interbank lending declined. After the crisis, banks lent unsecured to each other far less frequently, so the hypothetical became more hypothetical. Submitters were estimating the cost of transactions that were barely occurring, which left wide latitude that could not be checked against anything.
The Replacement
Regulators concluded the benchmark could not be repaired and it has been replaced with rates built on observed transactions rather than estimates.
The successors reference actual overnight borrowing in large, liquid markets, so the rate is a measurement rather than a survey. That removes the discretion that made manipulation possible.
The tradeoff is that these rates are overnight and secured, so they do not embed bank credit risk the way Libor nominally did. Contracts referencing them require spread adjustments to replicate what Libor was meant to capture, and the transition across existing contracts took years.
The Broader Point
The general lesson concerns any benchmark, index, or valuation produced by parties with a stake in the outcome. Appraisals commissioned by a seller, valuations of illiquid assets performed by the fund holding them, and ratings paid for by issuers all share the structure.
The question to ask about any reference number is who produced it, what verifies it, and whether the producer benefits from a particular answer.
The Bottom Line
Libor asked banks to estimate their own borrowing costs and used the answers to price the world. Benchmarks built on opinion rather than transactions carry manipulation risk by construction.