The Unravelling of UK Rate-Rigging Prosecutions: SFO Doctrine vs. Court of Appeal Ruling
The quashing of convictions against five former Barclays traders by the London Court of Appeal represents a definitive repudiation of the Serious Fraud Office's (SFO) original theory of benchmark prosecution. For over a decade, UK prosecutors argued that any consideration of a bank's own commercial trading positions when formulating an interbank rate submission (such as LIBOR or EURIBOR) constituted criminal conspiracy to defraud.
Under the British Bankers' Association (BBA) definition, banks were asked: “At what rate could you borrow funds, were you to do so by asking for and then accepting inter-bank offers in reasonable market size just prior to 11.00am?” Crucially, in money markets, borrowing costs are not an exact scalar; they are an interval characterized by credit tiering, counterparty appetite, time of day, and bid-offer spreads.
Maintained that submitters had a strict obligation to submit a purely objective rate, and that any trader request or commercial preference rendered the submission dishonest per se.
Recognized that when multiple rates accurately reflect legitimate borrowing possibilities within the permissible market range, choosing a rate that favors internal books is not inherently dishonest under the Ghosh / Ivey tests.
Mechanics of Interquartile Trimmed Mean Resiliency
A central quantitative question in benchmark litigation is whether individual submissions possessed the mathematical capacity to perturb the published fixing. As simulated in the workbench above:
- Rank Invariance: If a submitter raises their rate from 5.38% to 5.42%, but four other banks have already submitted rates higher than 5.42%, the target bank's submission is placed in the top quartile and eliminated from calculation entirely.
- Trimmed Damping: When the rate falls within the interior 8 banks (for a 16-bank panel), its contribution to the final fixing is diluted by 1/8th (12.5%). A 1-basis-point (0.01%) adjustment by a single bank moves the published benchmark by exactly 0.125 basis points (0.00125%).
- Counterfactual Zero-Impact: In dozens of historical instances cited by appellate defense teams, internal requests to “nudge” submissions had zero mathematical effect on the final index because the submission was either trimmed out or offset by independent moves among competitors.
Frequently Asked Questions on Benchmark Litigation & Appeals
Why were the five Barclays traders' convictions quashed in London?
The Court of Appeal found that the jury directions in the original trials were legally flawed. The trial judges had directed juries that it was unlawful as a matter of law to take commercial interests into account when setting LIBOR. The appellate court clarified that if the rate submitted was an honest estimate within the range of genuine borrowing rates available to the bank, taking commercial advantage was not an act of fraud.
How does the trimmed-mean mechanism protect against benchmark manipulation?
Under BBA rules for 16-bank panels, the four highest and four lowest submissions are automatically discarded before averaging the remaining eight. This eliminates aggressive outliers. To artificially shift the fixing when a rate is trimmed requires collusion across multiple institutions to skew the boundary of the middle 50% quintile.
What is the difference between permissible commercial discretion and criminal dishonesty?
Under current English criminal jurisprudence (applying Ivey v Genting Casinos), dishonesty is judged by the standards of ordinary decent people. Submitting a rate outside the genuine range of rates at which the bank could borrow, or fabricating nonexistent transactions to spoof the panel, remains fraud. Selecting an advantageous point within a legitimate range where liquidity actually exists is not criminal.
How have financial benchmarks evolved since the LIBOR scandal?
Global regulators have phased out subjective quote-based panels (like LIBOR) in favor of nearly risk-free, transaction-based overnight reference rates (e.g., SOFR in the US, SONIA in the UK, and €STR in the Eurozone). These indices are calculated purely from observed overnight repo or unsecured transactions rather than human bank estimates.