← All Projects

Network Analysis · Ownership Structures · AML Typologies

Art Markets & the PEP Problem

The art market's opacity is structural, not incidental — shell companies, freeports, and opaque intermediary chains create documented pathways for value movement that rarely trigger standard financial monitoring. This project reconstructs those pathways analytically, tracing where ICIJ offshore entity data, public auction records, and PEP networks intersect.

R· igraph Python · pandas ICIJ Offshore Leaks OpenSanctions Network Analysis FATF Typologies In Progress ◎

⬡ Entity Intersection Table ICIJ entities with traceable auction market presence, by jurisdiction & entity type

◈ Network Graph Ownership & intermediary structure around flagged nodes, betweenness centrality surfaced

◇ Typology Mapping FATF indicator annotations — analytically rigorous and compliance-literate

Project Progress

Data Assembly

Analysis

Published

I

The structural problem with art markets

Art is unusual as an asset class: it is portable, its value is subjective and difficult to verify externally, and its ownership has historically been opaque by design. Until the EU's 5th Anti-Money Laundering Directive (2020) and analogous national measures, auction houses and art dealers were largely outside the scope of AML frameworks that applied to banks and financial institutions. That gap closed late — and enforcement has been uneven.

FATF's 2023 guidance on art market money laundering identifies the core structural risk: the layering stage in particular benefits from the art market's tolerance for anonymous or nominee ownership, the use of freeport storage to hold assets outside normal customs and tax visibility, and the presence of sophisticated intermediary chains — private dealers, advisors, and holding structures — that place distance between the ultimate beneficial owner and any transaction record. These are not theoretical vulnerabilities. The Panama and Pandora Papers document specific instances of offshore structures used to acquire high-value art, with ownership deliberately routed through shell companies in low-transparency jurisdictions.

The analytical interest here is not simply "art is used for laundering." It is in the operational mechanics: how ownership structures and intermediary chains create observable traces across public datasets that are rarely put together in a single analytical view.

II

Analytical Question

Among entities and officers appearing in the ICIJ Offshore Leaks database, which demonstrate traceable participation — direct or through linked intermediaries — in major auction house sale records? And among those intersections, what structural characteristics emerge: which jurisdictions concentrate, which entity types recur, and which intermediaries bridge otherwise disconnected clusters?

A secondary question runs alongside it: where ICIJ-linked entities also appear in the OpenSanctions PEP registry, does the network structure around those nodes show features consistent with FATF's documented art market typology indicators — specifically, jurisdictional layering, nominee ownership, and freeport geography?

III

Analytical Architecture

The project builds across 3 linked analytical layers, each producing a distinct output that feeds into the next.

Layer 1 — Entity matching. Cross-reference ICIJ Offshore Leaks entity and officer names against publicly available auction house sale records using Python (pandas, fuzzy matching). The goal is a structured table of flagged intersections: entity name, ICIJ jurisdiction, auction house, sale year, and match confidence level. Scale matters less than analytical rigour here — even a small, well-validated intersection set is more credible than a large, noisy one.

Layer 2 — Network construction. Build a directed network graph in R using igraph, where nodes are entities (persons, shell companies, auction houses, jurisdictions, freeports) and edges are documented relationships (ownership, shared officer, sole participation, geographic linkage, intermediary role). The primary analytical output is this network — visualised to surface structural features: which nodes have high betweenness centrality (bridge entities), which jurisdictions cluster, and where the graph becomes locally dense in ways consistent with layering structures.

Layer 3 — Typology mapping. For each flagged cluster or structurally significant node, annotate against FATF's 2023 art market ML typology indicators. This is the interpretive layer that connects the graph findings to regulatory vocabulary: not a legal accusation, but a structured analytical mapping of what the data makes visible and which CDD or EDD provisions it would activate in a compliant infrastructure.

◈

Network Graph

Ownership & intermediary structure — ICIJ entities, auction houses, jurisdictions, freeport nodes. In progress.

Data Sources

01

ICIJ Offshore Leaks Database

Downloadable CSV — entities, officers, intermediaries, addresses, jurisdictions from Panama Papers, Pandora Papers, and related investigations. Primary dataset for entity matching and network construction.

02

Christie's & Sotheby's Public Sale Records

Publicly available lot-level sale results — lot title, hammer price, sale date, buyer/seller region. Used for entity matching against ICIJ records and for price anomaly context.

03

OpenSanctions PEP Registry

Structured, downloadable dataset of politically exposed persons across jurisdictions — name, role, linked entities, jurisdiction. Used for secondary cross-referencing against ICIJ-matched nodes.

04

FATF Art Market Guidance (2023)

Primary regulatory reference for typology mapping — specific ML indicators for auction houses, private dealers, freeport storage, and PEP-linked art market participation.

IV

What the output will show

The finished project delivers 3 outputs. A validated entity intersection table — ICIJ entities with traceable auction market presence, annotated by jurisdiction and entity type. A network graph visualising the ownership and intermediary structure around flagged nodes, with betweenness centrality surfacing the bridging entities that connect otherwise separate clusters. And a structured typology annotation mapping each significant finding to FATF indicator language — so the output is legible not just analytically but in the vocabulary of a compliance practitioner.

The goal is not to identify money launderers. It is to demonstrate what becomes analytically visible when you connect public datasets that are rarely read together — and to practice the cross-source, ownership-tracing analysis that underlies real financial crime intelligence work.

⬡

Entity Intersection Table

ICIJ entity × auction record matches — name, jurisdiction, sale year, match confidence. In progress.

✦

Analysis in progress. Entity intersection table, network graph, and typology mapping will be published here as outputs are completed.