Beyond Compliance: Evidence-Driven International Rulemaking

About the Author: Bio: Juan Carlos Portilla Jaimes is a leading scholar and educator in the field of international financial law. As Professor at Universidad de La Sabana’s Law School, he designs and teaches graduate‐level courses on cross-border banking regulation, anti-money laundering frameworks, asset recovery, and the intersection of financial crime and human rights. His classroom emphasizes case-based learning, comparative analysis, and policy development tailored to emerging global challenges.

An acclaimed author, Professor Portilla Jaimes penned "Transnational Financial Crime: An Unended Battle for International Law," a groundbreaking monograph. This work traces the evolution of multilateral legal instruments and exposes gaps in enforcement regimes. His rigorous research on AML and international law gained recognition in premier peer-reviewed such as the Harvard International Law Journal and the Cambridge International Law Journal.

Beyond academia, he serves as a consultant for intergovernmental bodies and non-profit coalitions seeking to strengthen anti-corruption protocols and promote financial integrity across jurisdictions. He regularly delivers lectures at institutions in Latin America, North America, and Europe, and at the American Society of International Law. His work bridges doctrinal scholarship and practical reform, equipping the next generation of lawyers and policymakers to advance the rule of law in a globalized financial system.

Introduction

The Financial Action Task Force (FATF) has long been the lodestar of global anti–money‑laundering and counter‑terrorist‑financing (AML/CFT) policy. Its 40 Recommendations set out standards designed to prevent and reduce money‑laundering, terrorist financing, and financing that facilitates the proliferation of weapons of mass destruction. For the purposes of this op‑ed, money‑laundering, terrorist financing, and financing the proliferation of weapons of mass destruction are grouped together under the umbrella term transnational financial crime. 

FATF Member States are expected to incorporate the FATF 40 Recommendations into their domestic legal frameworks for compliance purposes, with the goal of equipping national authorities with the legal tools to reduce transnational financial crime. Yet, the prevailing compliance paradigm—measuring success by the formal adoption of the FATF 40 Recommendations—has never been tested against the central question: Does compliance with the FATF 40 Recommendations actually reduce transnational financial crime? Complicating any attempt to answer this inquiry, FATF and states still rely on an outdated baseline—the United Nations’ 2009 estimate that criminals laundered roughly US$1.6 trillion—which no longer reflects a transformed global financial and technological landscape.

Accordingly, this op‑ed argues that the assumption that state compliance with the FATF 40 Recommendations causally reduces transnational financial crime should be tested using an AI‑driven global data system combined with rigorous regression designs. If causal evidence fails to support the hypothesis, the FATF framework requires substantive review and, ultimately, reform. The stakes are high: If compliance is reduced to the mere formal adoption of international standards, the international system risks squandering political capital and resources while transnational organized crime continues to exploit persistent gaps.

This article begins by arguing that the prevailing compliance paradigm has reached its limits. It then contends that the widely cited 2009 UN estimates continue to misguide contemporary international rulemaking. In response, it advances an evidence‑driven alternative grounded in the use of artificial intelligence to generate reliable data—specifically, estimates of the share of global gross domestic product laundered each year—and causal‑inference methodologies. The article further explains the policy relevance of this proposal, addresses foreseeable objections, and sets out the foundational principles of a new school of thought in international law. If the thesis developed here is correct, it could reshape how international organizations evaluate the real impact of the purposes and objectives embedded in treaties and international standards across all branches of international law.

Why the Compliance Paradigm is no Longer Enough:

While compliance focuses on formal adoption of treaties and international standards, evidence-based international rulemaking offers a clearer lens to global policymaking and the assessment of its real impact. States comply with international law when it serves their self-interests. Enforcement mechanisms such as trade benefits, sanctions, and military intervention reinforce compliance, as do technical and financial support programs. Chayes and Chayes contend that noncompliance often stems from practical constraints rather than defiance: poor planning, unclear treaty language, and limited institutional capacity. Social punishments reinforce compliance. As Johnston observes, international conferences and public reports, such as the FATF mutual evaluation process, create peer scrutiny that incentives states to adopt formal reforms. These evaluations, however, lack a data-driven, outcome-focused framework. Hence, it remains unclear whether the FATF mutual evaluations reports actually curb money laundering or merely promote technical adherence. 

Martin argues that focusing on compliance overlooks how international law changes state behavior, so research must distinguish mere adherence from real impact. Validating law through outcomes—what Martin and Simmons call “regime effectiveness”—has now become central to international law and international relations analysis. Despite this demand, most assessments of international law depend on assumptions or “anecdata.” Because systematic, data-driven compliance datasets remain scarce, assumptions made on anecdata are unreliable. EU scholars likewise highlight the absence of systematic policymaking metrics. As a result, global governance institutions lack systematic impact assessments, and there is no robust data to link compliance to anti-money laundering regulations’ effectiveness.

The compliance paradigm is useful—administrable, comparable, and politically visible—but counting laws and ticked boxes is an input, not the goal. FATF’s purpose is to reduce transnational financial crime and harm, not to devote efforts to the compliance paradigm.

The 2009 United Nations (UN) Report on Money‑Laundering: How Outdated Data Misguides International Rulemaking

The UN presented an estimate that criminals laundered roughly US$1.6 trillion in 2009, representing a substantial share of global gross domestic product and signaling the scale of transnational financial crime. The estimate was intended as a wake‑up call to international organizations to prioritize AML/CFT efforts. The UN report was derived from combining multiple estimation techniques and data sources available at the time, including macroeconomic indicators and case‑level information. The UN report relied on heterogeneous data, strong assumptions about the share of proceeds laundered, and extrapolations across jurisdictions with widely varying reporting quality. The figure reflects the 2009 financial and technological environment; it does not account for subsequent structural changes in payments, fintech, virtual assets, or criminal tactics.

Why is the estimate now problematic for international rulemaking? Since 2009, payment rails, digital finance, and cryptocurrencies have changed transaction speed, anonymity, and cross‑border flows. Criminals now use layering, trade‑based laundering, and decentralized platforms, diversifying channels and undermining the relevance of a single historical aggregate. Relying on the 2009 baseline therefore risks misallocating resources, privileging formal compliance with international standards over substantive outcomes, and overlooking emerging vulnerabilities. International lawmakers should treat the 2009 figure as historical context and commission a new, transparent global estimate using modern data and methods. They should adopt an evidence‑first agenda that combines AI‑assisted anomaly detection with causal‑inference evaluations of the FATF 40 Recommendations.

An Evidence‑First Alternative: AI + Causal Inference

The remedy is straightforward: complement box‑checking—formal adoption of treaties and international standards—with an outcomes‑oriented evaluation built on two pillars. The first is the use of AI to generate reliable data because traditional statistical methods cannot ingest, harmonize, and analyze the vast, heterogeneous, and fast‑moving financial, legal, and operational datasets required to estimate global illicit flows. Machine‑learning models also enable anomaly detection, cross‑jurisdictional pattern recognition, and real‑time signal extraction at a scale and speed that manual or conventional econometric approaches cannot match. Reliable data should include estimates of the share of global gross domestic product laundered each year by harmonizing legal, institutional, and operational information with anonymized and aggregated financial signals. This process should operate under strict privacy, legal, and ethical safeguards and is subject to independent oversight. 

The second is the application of rigorous regression and causal‑inference designs that use that harmonized, time‑varying data to move from correlation to causation and determine whether compliance with the FATF 40 Recommendations actually reduces transnational financial crime. The causation model uses a single dependent variable (y), whose variation the model seeks to explain, and several independent variables (x₁, x₂, x₃) hypothesized to cause changes in the single dependent variable (y). In compact form: y = f(x1, x2, x3…). The dependent variable would be triangulated through estimated illicit flows as a share of global gross domestic product, suspicious‑transaction‑report conversion rates, asset‑recovery performance per capita, and validated enforcement outcomes. The independent variable would be a continuous FATF compliance index that blends technical‑compliance scores with effectiveness proxies, supplemented by additional controls—such as gross domestic product per capita, financial‑sector depth, and governance quality—to isolate the effect of FATF compliance in regression analysis and strengthen causal identification. With these controls in place, a continuous FATF compliance index should exhibit a measurable effect on the dependent variable, meaning that higher compliance should be associated with a decrease in estimated illicit flows as a share of global gross domestic product. In this framework, accordingly, AI does not replace causal inference; it supplies the high‑quality, harmonized data that makes credible causal identification possible.

FATF Case Study: Applying the Causation Model to a Rising Tide of Financial Crime

The FATF regime offers an instructive case study—especially amid the allegedly global surge in transnational financial crime—for applying the causation model presented in this op‑ed. Although the FATF 40 Recommendations set global AML/CFT standards, their real effect on money‑laundering remains untested. International lawmakers should run causal, regression‑based analyses that link FATF compliance scores to money‑laundering (share of gross domestic product) while controlling for income, enforcement quality, political stability, and corruption.

Why this Matters for Policy

If causal evidence confirms the hypothesis, the FATF should pivot from technical checklists to outcome‑weighted peer reviews. Resources should be concentrated on the Recommendations with the largest marginal effects—beneficial‑ownership transparency, cross‑border cooperation, and STR conversion mechanisms—rather than spreading capacity building thinly across all FATF 40 Recommendations. FATF mutual evaluations should publish not only technical ratings but also causal impact estimates and enforcement outcomes, creating a feedback loop that rewards measurable effectiveness.

If causal evidence falsifies the hypothesis, the implications are more radical but no less constructive. The FATF would need to redesign its standards to prioritize interventions that demonstrably reduce transnational financial crime. That could mean shifting emphasis from formal transposition to operational capabilities (real‑time information sharing, automated STR triage, cross‑border investigative task forces), reweighing Recommendations to reflect impact, and redesigning peer review to trigger targeted reforms rather than symbolic compliance. Either way, the AI‑data plus causal‑inference model has the potential to produce better governance: It aligns incentives, clarifies priorities, and holds the FATF accountable to its own stated purposes.

Objections and Responses

Some will object that illicit flows are inherently unobservable and that any empirical program will be too noisy to guide policy. However, modern econometrics and data science can extract signal from noise using multiple proxies, robust identification strategies, and transparent uncertainty reporting, producing policy‑relevant estimates. Relying on the outdated 2009 UN Report and analysis applying the compliance paradigm alone will allow policy gaps to persist. Privacy and sovereignty must be engineered in.

A New School of International Law

Beyond FATF, this model points to a broader methodological shift: A new school of international law that treats the compliance paradigm as an input rather than the end. International law’s legitimacy depends not only on formal adoption of treaties and international standards but on demonstrable public‑interest outcomes. An AI‑driven, causation‑focused approach can be applied across international human rights, international environmental law, international trade law, and development: Did treaty adoption reduce deforestation? Did compliance with human‑rights reporting reduce torture? The tools are the same, and the normative demand is consistent: international law must be judged by the results it produces.

Conclusion:

The FATF 40 Recommendations were necessary but not sufficient. FATF can no longer treat the 2009 UN Report on money laundering baseline as definitive or equate formal compliance with progress. An AI‑driven global data system combined with rigorous causal analysis can test whether compliance actually reduces transnational financial crime. International law without evidence is faith; compliance without impact is bureaucracy. The FATF—and international law more broadly—must move from box‑checking to proof. The future of global governance may depend on it.