Who Connects Global Aid? The Hidden Geometry of 10 Million Transactions

Paul X. McCarthy, Xian Gong, Marian-Andrei Rizoiu, Paolo Boldi · 2025

Read Who Connects Global Aid? The Hidden Geometry of 10 Million Transactions by Paul X. McCarthy, Xian Gong, Marian-Andrei Rizoiu, Paolo Boldi (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2512-17243.txt.

Introduction

Paul X McCarthy, Xian Gong, Marian-Andrei Rizoiu, and Paolo Boldi map how aid money really moves around the world. Aid is often told as a story of country giving to country, which hides the groups the cash passes through. Using 10.9 million records, they trace over one million payments worth about one trillion US dollars among 22,773 organisations.

Size, they find, is not the same as power in the network. Big donors such as USAID rank only as central as their partner counts suggest. Foundations, research groups, and delivery firms sit in the most central spots. Nearly every group links into one web, yet money flows down long one-way chains. Losing one major funder could strand many groups below it.

Catalog record

CREATORPaul X. McCarthy, Xian Gong, Marian-Andrei Rizoiu, Paolo Boldi
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE36 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-17243

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Who Connects Global Aid? The Hidden Geometry of 10 Million Transactions Paul X. McCarthy, Xian Gong, Marian-Andrei Rizoiu, Paolo Boldi Abstract The global aid system functions as a complex and evolving ecosystem; yet widespread understanding of its structure remains largely limited to aggregate volume flows. Here we map the network topology of global aid using a dataset of unprecedented scale: over 10 million transaction records connecting 2,456 publishing organisations across 230 countries between 1967 and 2025. We apply bipartite projection and dimensionality reduction to reveal the geometry of the system and unveil hidden patterns. This exposes distinct functional clusters that are otherwise sparsely connected. We find that while governments and multilateral agencies provide the primary resources, a small set of knowledge brokers provide the critical connectivity. Universities and research foundations specifically act as essential bridges between disparate islands of implementers and funders. We identify a core solar system of 25 central actors who drive this connectivity including unanticipated brokers like J-PAL and the Hewlett Foundation. These findings demonstrate that influence in the aid ecosystem flows through structural connectivity as much as financial volume. Our results provide a new framework for donors to identify strategic partners that accelerate coordination and evidence diffusion across the global network. Introduction The modern architecture of global aid was born in the aftermath of World War II. In its infancy, Global Aid was conceived as a state-led instrument of geopolitical reconstruction managed by the OECD Development Assistance Committee [ 19 ] . It was driven primarily by diplomatic and domestic political imperatives, with the allocation often determined by donor strategic interests rather than recipient need [ 14 , 4 , 32 ] . However, the late 20th and early 21st centuries witnessed a radical phase transition [ 43 ] . A triple revolution characterised by new goals, new instruments and an explosion of new actors has reshaped the landscape [ severino2010end ] . This includes the rise of non-DAC donors [ 29 ] , the proliferation of vertical funds and private foundations, and a shift toward the complex interdependence of the Sustainable Development Goals [ 33 ] . Despite the growing scale, the politics of measurement remain contentious [ 52 ] . High-level policy attempts to harmonise this sprawling system, such as the Paris Declaration [ 39 ] , have largely failed. Empirical analysis suggests that coordination efforts have not reduced proliferation [ 3 ] . This leaves the sector in a state of unmanaged complexity, where uncoordinated high volumes of aid can sometimes hinder economic growth [ 13 ] . Today, the global aid system functions as a complex and evolving ecosystem. Our analysis of the International Aid Transparency Initiative (IATI) registry reveals a diverse population of 2,456 distinct publishing organisations ranging from multilateral banks and government agencies to private sector contractors and academic institutes. These actors operate across 230 countries and territories and utilise a complex mix of financial instruments including standard grants, aid loans and equity investments (Extended Data Fig. 1 ). The sheer scale of this system is now captured by over 10 million transaction records logged between 1967 and 2025. This complexity has outpaced the ability of policy makers to monitor the system through traditional statistical aggregates. In this environment outcomes emerge not from top-down directives but from millions of decentralised interactions between funders, implementers and intermediaries, that must be analysed as an autonomous, largely self-governed system. Despite this structural complexity, the prevailing view of the aid sector remains one that ranks actors primarily by who spends the most. This perspective highlights the financial dominance of massive bilateral agencies and multilateral banks such as the World Bank or USAID. However, this volume-based view obscures the hidden geometry of the system. In a networked system, influence and efficiency are functions of connectivity and brokerage as much as financial weight. Focusing solely on volume masks the critical structural holes where coordination fails. It also hides the distinct islands of humanitarian and development activity that remain sparsely connected. To date, there has been no systemic map identifying the structural brokers responsible for bridging these functional and geographic clusters. Here we present the first global-scale topological map of the aid system derived from a computational analysis of the IATI transaction corpus. By applying bipartite network projection and dimensionality reduction we reveal distinct functional clusters beyond traditional donor-recipient dyads. We identify a solar system architecture defined by the 100 most central actors with 25 at its very core. Crucially, we find that while governments and multilaterals dominate financial volume, a small set of knowledge brokers occupy disproportionately high bridging positions. We specifically identify universities and research foundations as the essential connective tissue holding the ecosystem together. We highlight the role of unanticipated central connectors such as J-PAL and the Hewlett Foundation. These findings suggest that to de-risk investments and accelerate evidence diffusion donors must distinguish between doers who deliver volume and connectors who maintain the structural integrity of the network. Results The Topology of Global Aid To understand the physical footprint of the aid system we mapped the geographic distribution of over 10 million transaction records logged in the IATI registry between 1967 and 2025. Unlike traditional volume maps which highlight only a few massive recipients, the transaction density map (Figure 1 ) reveals a truly globalised system involving activity in 230 countries and territories. We observe a heavy long tail distribution. While a small number of nations like India and Bangladesh represent massive hubs of high-frequency activity the majority of the network consists of nations with moderate transaction volumes that are often invisible in aggregate financial reporting. This geographic topology suggests a system that is widespread but highly fragmented. Activity is not evenly distributed but clustered around specific regional corridors. For instance we observe distinct high-density belts in East Africa and South Asia contrasted with sparser connectivity across Central Asia and South America. This fragmentation raises the question of whether these geographic clusters are integrated into a single cohesive network or function as separate islands of aid. Recent network literature challenges the traditional volume hypothesis [ 8 , 11 , 41 ] , suggesting instead that a recipient’s integration into the global network is a stronger predictor of growth and treaty ratification than raw financial flows [ 48 , 49 ] . Figure 1: Global distribution of 10 million aid transactions from 1967 to 2025. Darker regions indicate higher transaction density on a logarithmic scale. This reveals the geographic footprint of the 2,456 publishing organisations in the International Aid Transparency Initiative (IATI) registry. While activity is global it is highly clustered around specific regional corridors in East Africa and South Asia with India and Bangladesh representing massive high-frequency hubs. The Hidden Geometry of Funding and Implementation To test for this structural integration, we applied Uniform Manifold Approximation and Projection (UMAP) to the network embeddings of all 2,456 provider organisations. To derive these embeddings, we first constructed a bipartite graph consisting of nodes representing organisations of various types (e.g., Government, NGO, Private Sector) that either provide or receive fundings. The connections between these nodes are established by financial aid transactions, that is, specifically connecting organisations that provide funding to those that receive it. We then processed

Read the complete text