Benchmark manipulation as private public infrastructure failure

Benchmarks become system assets when private inputs set public contract terms

The paper treats LIBOR, FX fixes, ISDAFIX, and credit ratings as shared price infrastructure. The failure appears when the actor who can move the reference object also has benchmark-linked payoff exposure and the record cannot reconstruct the decision.

• Game g has actor A, counterparty B, system S, payoff Π, system welfare WS, and record R.
• The system cost sits outside the private ledger: trust decay, monitoring cost, capital misallocation, transition burden, and supervisory opacity.
• Scope is narrow: the claim applies only when public reliance, private influence, conflict exposure, and record incompleteness hold together.

A private benchmark can carry $350 trillion of public reliance

The paper’s infrastructure mismatch is scale. Benchmarks coordinate contracts, valuation systems, hedges, and regulation, while submissions, fix-window trading, and rating committees remain controlled by private actors.

• Panel banks submitted LIBOR borrowing estimates while holding positions linked to the published rate.
• Dealer desks traded around the WM/Reuters 4pm London Fix while client orders and benchmark-linked flows were concentrated in the window.
• Rating agencies became benchmark-like when mandates, capital rules, and structured-finance contracts relied on private grades.

The theorem needs all four conditions to create a positive welfare floor

The Benchmark Infrastructure Floor Theorem is a local result about benchmark regimes. It says no private payoff rule limited to direct parties can guarantee zero welfare loss when the four axioms jointly hold.

• Axiom 1: public reliance r(B) > 0 through contracts, valuation, regulation, or mandates.
• Axiom 2: private influence over the reference object through submissions, trades, ratings, methodology choices, or exceptions.
• Axiom 3: conflict exposure c(A,B) > 0 because the actor’s payoff can rise when the benchmark moves.
• Axiom 4: record incompleteness q(R) < 1 because an external reviewer cannot fully reconstruct evidence, conflicts, review, and infrastructure effects.

Removing one axiom can collapse the benchmark infrastructure floor

The paper verifies axiom independence through countermodel analysis. The welfare floor is not caused by private administration alone, public use alone, or conflict alone; it comes from their combination with a missing reconstruction record.

• No public reliance: a private estimate can be wrong without becoming public infrastructure harm.
• No private influence: a reference object may be widely used but not manipulable by the actor.
• No conflict exposure: influence without benchmark-linked payoff does not generate the same manipulation incentive.
• Complete record: q(R) = 1 changes the identification problem because evidence, conflict, alternatives, and review become reconstructable.

LIBOR shows estimate-based submissions, panel-bank conflicts, and crisis-period bias

LIBOR was vulnerable because the published rate depended on bank submissions rather than transaction-anchored borrowing data. The paper uses empirical estimates as distortion evidence and keeps private gains separate from welfare loss.

• Youle (2014): approximately 8 bps downward bias during 2007-2009.
• Bonaldi (2016): 23 bps average deviation at the worst of the financial crisis.
• Kops and Conrad (2015): 29.3-33.1B cumulative gains for panel banks from 1999 through 2010.
• Penalties are not counted as welfare loss; they establish conduct and enforcement record.

FX fix manipulation used concentrated trading around the WMR fixing window

The FX module is not a repeat of LIBOR. The mechanism is fix-window trading rather than quote submission: traders could coordinate order handling and trading around a benchmark time used for valuation and execution.

• Evans (2018) documents dealer profits from WMR fix-window manipulation.
• Ito and Yamada (2016) show governance reform changed trading costs and risks around fixing.
• El Mouaaouy (2015) identifies FX trading patterns associated with manipulation hot spots.
• The welfare channel is execution trust, deterred participation, monitoring, and spread effects, not gross daily FX volume.

ISDAFIX and ratings extend the record problem through different mechanisms

The paper groups ISDAFIX and credit ratings with LIBOR and FX fixes because each creates a private input to a public reference object, not because the conduct is identical.

• ISDAFIX involved swap-rate submissions and benchmark influence in derivatives markets.
• Credit ratings become benchmark-like when NRSRO grades feed mandates, capital treatment, structured-finance contracts, and investor constraints.
• The issuer-pays model creates conflict exposure because the agency’s revenue comes from issuers while the grade is used by the market.
• The paper uses no aggregate harm number for credit ratings.

The paper counts real-resource costs, not transfers, penalties, or notional exposure

Cash-flow shifts from a manipulated benchmark begin as transfers. Welfare loss enters only when the benchmark distortion changes real decisions or consumes real resources.

• C1 distorted hedging, allocative inefficiency, and deterred trades: MC median $3.5B/yr.
• C2 trust, transition, monitoring, and compliance rebuilds: MC median $2.5B/yr.
• C3 regulatory remediation real costs: MC median $1.0B/yr.
• C4 litigation real costs: MC median 0.5B/yr; C5 market integrity premium: MC median 0.6B/yr.

The calibrated regime loses 8.4B/yr against 4.0B/yr of private payoff

The System Asset Pricing Model estimates β_W as the ratio of system welfare cost to private payoff. The paper reports 100,000 Monte Carlo draws with seed 42 and correlation ρ = 0.3.

• Median β_W = 2.1 with 90% CI [1.1, 3.9].
• P(β_W < 1) = 3.7%, so welfare improvement is possible only in extreme draws.
• System-adjusted payoff ΠSA = -$4.4B/yr.
• Classification remains Hollow Win (Type II) across lognormal, normal, and triangular families; CV = 1.9%.

Field 17 records infrastructure effects before submissions, fixes, ratings, and exceptions

Decision Accounting is the paper’s reconstruction layer. It does not promise prevention; it makes benchmark decisions auditable before system harm has to be inferred from settlements and after-the-fact screens.

• Field 17 captures infrastructure effects beyond direct parties: integrity, liquidity, allocative reliability, and supervisory visibility.
• A benchmark record should identify actor, evidence, conflict, alternative, reviewer, escalation, and cost bearer.
• The target record-completeness level reported in the findings is 0.90.
• The standard is designed to fit IOSCO, EU BMR, and FSB reform architectures.

Transaction anchoring lowers the floor only if records become reconstructable

The paper’s remedy changes game g into g1 through benchmark architecture, not only stronger punishment after harm. LIBOR transition is treated as real-world game-change evidence, but transaction anchoring alone is incomplete.

• LIBOR to SOFR, SONIA, €STR, SARON, and TONA reduces reliance on expert judgment by anchoring rates to transactions or risk-free rate architecture.
• Governance separation reduces the actor’s ability to influence the reference object while holding benchmark-linked exposure.
• Decision Accounting records target the remaining gap: methodology exceptions, fallback choices, rating actions, and review trails.
• The policy test is whether the regime improves r(B), c(A,B), q(R), and κ in the floor expression β_W ≥ r(B) × c(A,B) × (1 - q(R)) × κ.