Benchmark Manipulation

1 Introduction

Benchmark manipulation is the distortion of a widely used reference rate so that it no longer reflects honest market conditions. Many benchmarks are built from inputs supplied by a panel of banks, or measured during a short daily "fixing" window. In the LIBOR scandal, traders asked colleagues to submit false borrowing-cost estimates to move the published rate in a direction that benefited their derivatives positions. In the foreign exchange "fixing" scandal, dealers coordinated to push large orders into the brief window used to calculate the WM/Reuters fix, a practice nicknamed "banging the close". Because these rates price trillions of dollars of loans and derivatives, even a tiny shift moves enormous sums. Global regulators ultimately imposed billions of dollars in fines across both scandals.

Trillions

In contracts priced off these benchmarks

$9B+

Combined LIBOR and FX fixing fines

60 sec

Typical length of an FX fixing window

2 Interactive Benchmark Fix Simulation

Phase: Honest Submissions

Panel Bank Submissions

The Fix and the Trader Position

Step 1 - Honest Submissions: Each panel bank reports its true estimate. The benchmark is a trimmed mean: the highest and lowest submissions are dropped, and the rest are averaged. The fix sits at the genuine market rate.

3 Detailed Analysis

Two Ways to Rig a Benchmark

Survey Manipulation (LIBOR, EURIBOR)

The benchmark is built from estimates that panel banks submit. Submitters report false figures, often at the request of derivatives traders, to nudge the trimmed-mean rate up or down.

Window Manipulation (FX Fixing)

The benchmark is measured during a short window. Dealers cluster large orders into that window, a tactic called "banging the fix", to push the observed price where their book profits.

Why a Tiny Move Matters

Benchmarks like LIBOR and the daily WM/Reuters FX fix anchor the price of an immense volume of loans, mortgages, and derivatives. A shift of even a fraction of a basis point is trivial on a single trade, but when it is applied across a multi-trillion-dollar pool of contracts the transfer of value is vast. The trimmed-mean design, which drops the highest and lowest submissions, is meant to blunt a single outlier. It fails when several contributors coordinate, because their pushed values survive the trim and drag the average with them.

Detection Methodology

Investigators compare a bank's submissions against its actual funding transactions and against the wider market, flagging inputs that drift away from observable reality. They reconstruct chat logs and messages to find requests to move the rate, and they correlate the direction of submissions or fixing-window orders with the bank's derivatives positions. Statistical tests look for clustering of trades in the seconds around a fix and for submissions that repeatedly sit just inside the trim threshold. Modern surveillance ties communications, order timing, and position data together to expose coordination across institutions.

Red Flags

Submissions that diverge from a bank's own funding costs or from the observable market
Communications requesting a higher or lower input that align with a trading position
Order flow clustering in the brief seconds around a daily fixing window
Coordination across multiple contributor banks moving submissions in the same direction