News

How China Collaterizes: New Report Reveals Hidden Structures in Global Lending

26.06.2025

Today, an international team of lawyers, economists, and political scientists launched a report that sheds new light on the secured lending practices of Chinese creditors in low- and middle-income countries and highlights the urgent need for more transparency in collateralized lending and borrowing. The report shows how these creditors secure priority access to cash flows and liquid assets that can be used to repay risky loans to emerging markets and developing economies (EMDEs). In a typical transaction, debtors promise to route their principal commodity export revenues through overseas bank accounts that remain out of public sight and largely beyond their control until the debts are repaid. The cash balances in these accounts, mostly located in China and controlled by the lenders, can be very large; in low-income, commodity-exporting countries, they average more than 20% of annual public debt service to all external creditors.

The How China Collateralizes (HCC) report, by Anna Gelpern, Omar Haddad, Sebastian Horn, Paulina Kintzinger, Brad Parks and Christoph Trebesch, is the first comprehensive analysis of the secured lending practices of Chinese creditors in EMDEs. It draws upon in-depth case studies and a new dataset of collateralized public and publicly guaranteed (PPG) loans from Chinese state-owned institutions to EMDEs. In total, the dataset captures 620 collateralized PPG debt transactions worth $418 billion in constant 2021 USD over a 22-year period. The report and the dataset are available for download.

Today, in conjunction with the release of How China Collateralizes, AidData has published a large cache of debt contracts—including loan, escrow account, mortgage, and debt restructuring agreements—between Chinese creditors and their overseas borrowers. In total, it has published 371 contracts between 19 Chinese creditors and 155 borrowers from 60 countries in an online repository. Digitized copies of the contracts can be accessed and searched by lender, borrower, sector, and contract clause at http://china-contracts.aiddata.org/.

Revenues stuck in offshore bank accounts reduce the assets available to pay unsecured creditors and government expenses

The authors of the report made a number of unexpected findings.

“We were surprised to find that almost half of China’s PPG lending portfolio, or nearly $420 billion across 57 countries, is effectively collateralized—mostly with deposits in bank accounts abroad,” said Christoph Trebesch of the Kiel Institute for the World Economy. “As security, Chinese lenders strongly prefer liquid assets—in particular, cash deposits in bank accounts located in China. They also want visibility and control over revenue streams.”

Foreign currency revenues deposited in bank accounts controlled by Chinese lenders secure approximately 80% of the collateralized lending volume in the new dataset. A typical security package supporting a Chinese PPG loan includes one or more restricted (escrow) accounts at banks located in China, funded by revenues from the borrowing country, bolstered by contract and property rights in the cash flows. In many cases, the deposit account is at the creditor bank, which gives them a high level of control over some of the borrower’s core revenue streams, as well as set-off rights under Chinese law. “It’s rare that Chinese creditors secure their exposures with local currency revenues and bank accounts in the borrowing country, or with physical assets such as land or equipment,” added Trebesch.

Unrelated Revenues, Elevated Risks

The research team also found that collateral is often unrelated to the stated purpose of the loan. “We see Chinese lenders expanding and adapting standard market tools to make exceptionally risky loans safer,” said Anna Gelpern, a Georgetown Law Professor and Nonresident Senior Fellow at the Peterson Institute for International Economics. “Instead of relying on infrastructure project assets and future revenues, which may never materialize, they seek access to established export proceeds. Exporters commit to route these proceeds through offshore bank accounts over the life of the loan, which gives creditors leverage in the relationship as well as a source of repayment.” The report notes that the World Bank and the IMF have recently raised concerns about “collateralization involving unrelated assets or revenues” and warned that it is “likely to create problems.”

Oil, Gas, Gold, Cocoa: The Commodities Behind the Collateral

According to the authors, commodity revenue sources vary by borrowing country, but typically draw on that country’s leading commodity export: oil in Angola, Iraq, Russia, Sudan, South Sudan, Equatorial Guinea, the Republic of the Congo, Brazil, and Venezuela; gas in Indonesia, Myanmar, and Turkmenistan; gold in Kazakhstan; copper and cobalt in the Democratic Republic of the Congo; bauxite in Guinea; platinum and tobacco in Zimbabwe; cocoa in Ghana; and sesame in Ethiopia. Oil proceeds dominate, accounting for 79% of the commodity-backed lending volume in the dataset.

“Our research reveals a previously undocumented pattern of revenue ring-fencing, where a significant share of commodity export receipts never reaches the exporting countries,” said Brad Parks, Executive Director of William & Mary’s AidData research lab. “When the revenues from a country’s principal source of foreign currency secure its debts to a single creditor, unsecured creditors are effectively subordinated and the risk of a destructive collateral ‘arms race’ increases. Multilateral development banks should seek new ways to improve fiscal governance and ensure compliance with negative pledge commitments among vulnerable commodity exporters with a poor track record of fiscal management.”

Collateral without a Contract: The Rise of Quasi-Security

The new study also reveals previously unknown details about collateralization mechanisms in China’s PPG lending portfolio in the developing world. Chinese creditors are far more likely to obtain de facto control over revenue streams and cash holdings (“quasi-collateral”) than formal security interests (liens, pledges, charges, or assignments) in the assets. Quasi-collateral can have the same economic effect as a formal security interest, but it comes with few or no public notice requirements for debtors or creditors.

Over time, the sums that accumulate in offshore accounts to secure public infrastructure debts can be very large, up to billions of US dollars. Political oversight institutions in borrowing countries—such as supreme audit institutions and public accounts committees within parliamentary bodies—have found it difficult to monitor encumbered revenue streams and cash holdings in China.

The implications for EMDEs are far-reaching. “These transactions put those tasked with fiscal governance and debt crisis management in borrower countries in difficult situations,” said Paulina Kintzinger of the Kiel Institute for the World Economy. “Secrecy makes it more difficult to untangle creditor claims in times of distress and default, and can undermine both fiscal autonomy and macroeconomic surveillance.”

Participants in the Global Sovereign Debt Roundtable recently noted that “undisclosed collateralized debt poses particular challenges” for sovereign debt management and debt restructuring. They also agreed that secured and unsecured debt must be restructured on comparable terms.

One Pool, Many Debts: Cross-Collateralization Structures

According to the research team, Chinese lenders also use cash collateral pools that simultaneously secure multiple debts, creating a web of interconnected transactions and risk transmission pathways. A default on any one of these debts can trigger creditor enforcement against the shared collateral, irrespective of the payment status of other obligations, and reverberate across the government’s debt stock and beyond.

For nearly half of China’s collateralized PPG loans in EMDEs, the same asset or pool of assets secures more than one loan. The transaction structure fuses elements of pre-export finance, traditionally used to support commodity exporters, with public infrastructure finance. In the combined structure a pool of established export revenues serves as collateral to de-risk multiple unrelated domestic projects in developing countries.

According to Omar Haddad of Oxford University, “China Eximbank developed a replicable and scalable cross-collateralization structure in Angola. That model was subsequently refined and adapted by a diverse group of Chinese creditors in a wide array of PPG lending operations in Africa, Latin America, the Middle East, and Central Asia.”

Public debt is not public

It took the team of researchers nearly four years to make sense of the opaque and complex borrowing arrangements that are documented in the report. The obstacles that they faced were formidable. Despite recent advances in debt data sharing, few bilateral or commercial creditors publish their secured lending terms. Many seek confidentiality commitments from their borrowers, impeding disclosure. Chinese creditors are no exception. They do not publish detailed or comprehensive data about their collateralized PPG loan agreements with EMDE borrowers. Their security and escrow account agreements are even harder to obtain. Quasi-collateral structures present an additional challenge: revenue routing and restricted accounts are substantially easier to shield from public scrutiny than liens over physical assets. The same structures can also undermine fiscal autonomy, debt and revenue accountability, and macroeconomic surveillance.

The Authors

Anna Gelpern is a Scott K. Ginsburg Professor of Law and International Finance at Georgetown Law and a Nonresident Senior Fellow at the Peterson Institute for International Economics;
Omar Haddad is a lawyer and graduate student at Oxford University;
Sebastian Horn is a Professor of Economics at the University of Hamburg and the Kiel Institute for the World Economy;
Paulina Kintzinger is a PhD candidate at the Kiel Institute for the World Economy;
BradParks is the Executive Director of AidData and a Research Professor at William & Mary’s Global Research Institute;
Christoph Trebesch is a Professor of Economics at Kiel University and Director of the "International Finance and Macroeconomics” Research Area at the Kiel Institute for the World Economy.

The authors will be available for comment in the lead-up to the embargo lifting, as well as during and after the launch, through the media contacts below.

Data Collection Sources and Methods

To construct the How China Collateralizes (HCC) Dataset (Version 1.0), the research team first identified all collateralized PPG loan commitments in the 3.0 version of AidData’s Global Chinese Development Finance (GCDF) dataset. It then used AidData’s Tracking Underreported Financial Flows (TUFF) methodology to identify sources that document the collateralization arrangements of individual borrowings.

In total, the research team drew upon more than 12,000 sources to build the dataset, including the annual reports, financial statements, stock exchange filings, and bond prospectuses of borrowing institutions; IMF Article IV reports and World Bank-IMF debt sustainability analyses (DSAs); loan agreements, account agreements, mortgage agreements, share pledge agreements, account charge agreements, assignment of receivables agreements, deeds of security, and deeds of covenant; and official correspondence between lenders and borrowers; and media reports.

Each source was reviewed in full and a standardized, replicable, and transparent set of protocols was followed to capture the legal form and the economic function of the collateralization arrangements. The 1.0 version of the HCC dataset introduces 27 new variables that measure (1) sources and types of collateral; (2) cross-collateralization and cash collateral pooling; and (3) the use of escrow accounts, including the minimum cash balances that borrowers commit to maintain in the accounts and the actual cash holdings that borrowers maintain over time.

When currency swap borrowings are excluded, the dataset captures 620 PPG loan commitments worth $418 billion in constant 2021 USD for approved, active, and completed projects and activities from 31 Chinese state-owned creditors to 158 borrowers in 57 EMDEs between 2000 and 2021. The dataset does not capture suspended and cancelled loan commitments; projects and activities that Chinese state-owned creditors agreed in principle to finance but never resulted in formal loan commitments; or so-called “umbrella” agreements (such as master framework agreements) that sought to finance multiple projects and activities via subsidiary loans.