Turning Compliance Theatre into Real Control

Measuring the effectiveness of anti‑money laundering (AML) has become a ritual in itself. We obsess over what can be easily counted – suspicious activity reports (SARs), number of staff, training sessions delivered, technology spend – and then quietly treat these as proxies for success. Yet none of these metrics tell us the one thing that matters: are we actually reducing criminal revenues, constraining serious offenders, and improving the lives of those harmed by economic crime? (While some defend AML goals as broader deterrence or systemic integrity, the core test remains crime reduction.)​

The uncomfortable truth is that, three decades into the global AML experiment, we still reward visible effort over demonstrable impact.

The Mirage of AML “Effectiveness”

Global AML systems today are a textbook case of legal endogeneity: the regulated sector, consultants, and standard‑setters co‑produce the very definitions of “good practice” against which they are later judged. In that self‑referential world, two things dominate:

Inputs and processes – policies, procedures, systems, attestations.
Volumes – SAR counts, alerts, audits, training hours.

These are taken as evidence of effectiveness, even though they are at best tenuous proxies for any reduction in crime – with global asset recovery consistently estimated at under 1% of criminal proceeds (FATF/UNODC ranges). Meanwhile, the hard outcome data – meaningful investigations, convictions, asset recovery, and the dismantling of high‑impact criminal infrastructures – remain stubbornly weak. We continue to circulate huge, methodologically fragile estimates of global laundering in the trillions, while recovering only a tiny fraction of those proceeds (e.g., <0.1-1% per IMF/UNODC) and securing relatively few convictions.

This mismatch is not a minor technical flaw; it is a structural feature. AML has become a system that has grown more adept at generating symbols of vigilance – SARs, frameworks, certifications, ratings – while struggling to impose real friction on serious criminal activity.

Symbolic Compliance and the Comfort of Numbers

Why does this persist? Because symbolic compliance is politically and commercially convenient.

• Politicians can point to tough laws, national risk assessments, and glowing evaluation reports.
• Regulators can demonstrate “robust supervisory frameworks” and rising fine totals.​
• Financial institutions can showcase sophisticated systems, armies of compliance staff, and impressive SAR statistics.

Everyone gets to claim success without confronting the awkward question: if the system is so effective, why do we still see such a vast gap between estimated criminal proceeds and what is actually restrained, confiscated, and returned?​

The answer is that much AML activity is not designed to optimise crime reduction. It is designed to minimise institutional and personal risk: regulatory, reputational, and political. In this world:

• Producing large quantities of low‑value SARs is safer than producing fewer, better‑targeted reports.
• Investing in visible, generic tools is easier than challenging products, sectors, or trade corridors that are structurally attractive to criminals.
• Chasing “paper compliance” is more rewarding than challenging the system’s foundational assumptions.

The result is a system in which banks act as reassurance producers and regulators as reassurance consumers, while serious launderers continue undisturbed in the background.

TBML: The System’s Blind Spot

Nowhere is this illusion of control more apparent than in trade‑based money laundering (TBML). TBML is likely one of the most consequential laundering methods globally, yet it remains conceptually confused, empirically under‑examined, and operationally under‑policed.​

Historically, TBML has been framed as a niche problem of cross‑border documentary trade – letters of credit, discrepant documents, and classic mis‑invoicing. The reality, supported by case studies and empirical interviews, is far broader: a spectrum from crude trade abuse to sophisticated manipulation of domestic and cross‑border trade in goods and services, often blending licit and illicit activity.​

Crucially, the parts of trade that are easiest to monitor are not necessarily the parts criminals use most:

• Documentary trade finance receives intense compliance attention because it generates artefacts that fit neatly into existing AML controls.​
• Open‑account trade, service trade, domestic trade, and hybrid arrangements that move value through legitimate supply chains remain comparatively under‑scrutinised.

In the UK, this problem has been amplified post‑Brexit. The Government’s economic strategy explicitly seeks to expand trade with higher‑risk developing markets, even as law enforcement loses seamless access to EU‑level intelligence tools and already‑limited specialist capacity is stretched further.

My empirical research with senior TBML professionals in UK banks found a consistent pattern: TBML is formally recognised as a major risk in national narratives, but is operationally deprioritised by both banks and law enforcement. Interviewees highlighted:​

• Deficient skills, resources, and technology for TBML detection.
• A lack of strong regulatory stimulus to treat TBML as more than a rhetorical priority.​
• Structural reliance on under‑resourced law enforcement to define priorities and provide typologies, even though law enforcement itself lacks TBML capacity and incentives.​

In practice, some of the most vulnerable vectors banking professionals stressed are small and medium‑sized import–export firms trading with Asia, Latin America, and Sub‑Saharan Africa, often on open‑account terms. These are:

• Operationally hard to know well (limited relationship management, fragmented data, weak documentation).
• Politically sensitive to touch, given their role in trade and development narratives.
• Extremely attractive to TBML actors seeking to blend illicit value into legitimate flows.

When a sector is simultaneously high‑risk and high‑story‑value (jobs, exports, growth), the path of least resistance is symbolic oversight: enough controls to claim attention, not enough to materially disrupt the flows.

Technology: Tool or Theatre?

AML and TBML discussions are saturated with references to “advanced technology”: machine learning, network analytics, AI, natural language processing. On paper, these tools promise a step‑change in pattern recognition. In reality, their deployment is heavily constrained by regulatory conservatism (e.g., bias toward rules-based systems generating 90-99% false positives), vendor incentives favouring audit-friendly generics over TBML-specific models, and institutional risk aversion.​

The outcome is predictable. Many institutions operate expensive systems optimised to produce large volumes of alerts (e.g., transaction monitoring false positive rates often >95%), not to uncover supply‑chain manipulation involving domestic and multi‑jurisdictional trade. If technology is to support effectiveness rather than theatre, it must be paired with staff who have both domain expertise and the institutional mandate to use the insights to challenge business and client selection decisions. Without that, “AI in AML” remains a shiny layer on top of a structurally misaligned regime.

What an Effectiveness‑Focused Paradigm Would Really Look Like

If we take effectiveness seriously, several shifts are unavoidable.

1. Redefine the Metrics

We need to move from input and process metrics to outcome and harm‑based indicators. That means:

• Tracking the quality, not just the quantity, of financial intelligence (e.g., SARs that lead to actionable investigations/Dissemination rates >10-20%, per industry benchmarks).
• Systematically measuring confiscations, restraint orders, successful prosecutions, and the dismantling of high‑impact networks, not just “cases touched”.​
• Analysing how typologies and methods evolve in response to regulatory and enforcement actions, rather than declaring victory the moment a new rule is enacted.

Critically, these data should be public and granular enough to allow external scrutiny, not buried in internal dashboards and self‑congratulatory reports.

2. Realign Public and Private Incentives

For the public sector, effectiveness means shifting supervisory practice from box‑ticking to outcome‑oriented evaluation. Supervisors should:

• Reward institutions that can demonstrate meaningful contributions to crime control, even if their approach departs from the prevailing orthodoxy.
• Challenge institutions that hide behind volume metrics and generic tools while offering little evidence of real‑world impact.

For the private sector, it means accepting that “not being fined” is not the same as “being effective”. Banks should be incentivised, and socially expected, to:

• Prioritise high‑quality, collaborative intelligence over defensive, volume‑driven reporting.
• Invest in TBML‑relevant capabilities (data, expertise, cross‑functional teams) rather than merely scaling generic AML infrastructure.

3. Use Technology to Understand, Not Just Signal

Technology should be evaluated against a brutally simple test: does it measurably improve our understanding of and ability to intervene in high‑impact laundering, including TBML? If not, its role is performative.

Regulators have a central role here. They must be willing to:

• Engage with more complex analytical methods and reject dumbed-down explanations on how these work that sanitise complexity for bureaucratic comfort.
• Tolerate justified experimentation and iterative improvement, rather than enforcing frozen “best practice” that locks in today’s blind spots.

4. Reform How We Judge Countries – Including Ourselves

The Financial Action Task Force (FATF) mutual evaluation process has grown into a dense machinery of recommendations, interpretive notes, and methodologies. Yet even with its Immediate Outcomes focus since 2013, it still over‑emphasises formal alignment while under‑emphasising sustained crime outcomes (e.g., UK’s high technical scores vs persistent illicit finance conduit status).

Evaluations frequently:

• Over‑emphasise the existence and formal alignment of laws, institutions, and frameworks.
• Under‑emphasise whether those frameworks have produced sustained, demonstrable improvement in crime outcomes.

If the upcoming rounds of mutual evaluations is to escape the charge of ritualism, it must be prepared to:

• Confront the dissonance between strong technical scores and weak enforcement results in systemically important jurisdictions.
• Demand evidence of how AML regimes have affected actual laundering patterns, especially in structurally vulnerable areas like TBML.

For countries like the UK, which have received glowing evaluations while simultaneously being identified as major conduits for illicit finance, this will require real self‑scrutiny rather than defensive celebration.

The Moral Stakes of Getting This Wrong

TBML and other laundering methods are not abstract technical puzzles; they are enabling infrastructures for exploitation, corruption, environmental degradation, and violence. When we accept a system that prioritises symbolic AML over substantive disruption, we are implicitly accepting that these harms will continue, so long as they remain offstage and offshore.

The choice is not between “perfect AML” and “no AML”. It is between:

• A regime that optimises for appearances and institutional risk management.
• A regime that optimises for reducing criminal power and protecting human welfare, even when that creates discomfort for powerful economic and political interests.

If we are serious about effectiveness, then the configuration of rules, institutions, incentives, and technologies cannot be sacrosanct. It is precisely that configuration which needs to be on the table.

Until we are prepared to judge our AML systems by the damage they prevent – not the paperwork they generate – we will continue to live in an elaborate theatre of control, while trade‑based money laundering and other forms of illicit finance flourish in the wings.