Debt-Trap Diplomacy, Tested: $35B Due to China in 2025

The central finding of this report is that the debt-trap thesis, in the form that made it famous, does not survive contact with the loan data — but the problem it tried to name is real and larger than its critics admit. No documented case exists of Beijing foreclosing on a strategic asset because a sovereign borrower failed to pay ◈ Strong Evidence [7][8]. The port most often cited as proof, Hambantota in Sri Lanka, was leased in a deal that forgave no Chinese debt, and the cash went to Sri Lanka’s reserves [8][9]. Yet the contracts behind Chinese lending are engineered to put Beijing at the front of the repayment queue, to keep its debt out of collective restructurings and to keep the terms secret ✓ Established Fact [12][13]. The leverage exists. It simply runs through the balance sheet rather than through the harbour.

The scale is the first thing the popular debate gets wrong, and it gets it wrong in both directions. AidData’s November 2025 dataset — built over 36 months by 16 full-time researchers and 126 assistants — counts 30,133 projects financed by 1,193 Chinese state-owned lenders and donors, worth 2.2 trillion dollars between 2000 and 2023 [1][2]. Brad Parks, AidData’s executive director, says the portfolio is two to four times larger than earlier published estimates [2]. But the geography has shifted in a way the trap narrative never anticipated: more than 75% of the portfolio now supports upper-middle and high-income economies, 943 billion dollars has gone to 72 high-income countries, and the single largest recipient of official Chinese credit is the United States, at more than 200 billion dollars ✓ Established Fact [2].

For the poorest borrowers the direction of money has reversed. The Lowy Institute calculates that developing countries will pay 35 billion dollars in debt service to China in 2025, of which 22 billion comes from 75 of the world’s poorest and most vulnerable economies [3]. New Chinese lending peaked at around 50 billion dollars in 2016 and has fallen to roughly 7 billion a year since 2023; net flows to developing countries turned negative, at minus 34 billion dollars, in 2024 ✓ Established Fact [3]. Riley Duke, the report’s author, concludes that for the rest of this decade China will be more debt collector than banker to the developing world [3]. That is the structural fact from which every other argument in this report proceeds.

China’s weight as a creditor is now without modern precedent. In 54 of 120 developing countries with available data, debt service owed to China exceeds the combined payments owed to every Paris Club lender — the bloc that includes all the major Western bilateral creditors ✓ Established Fact [3]. China is the largest bilateral creditor in 53 countries, holds 26% of external bilateral debt worldwide and more than half in the poorest economies [3]. Lowy’s judgement is blunt: no single bilateral creditor has been responsible for such a large share of developing-country debt service in the past 50 years [3]. None of this proves a trap. It does establish that when a Belt and Road borrower gets into trouble, the creditor on the other side of the table is almost always Beijing.

That concentration arrives at the worst moment in the global rate cycle. Developing countries paid a record 1.4 trillion dollars in external debt service in 2023, with interest payments up by nearly a third to 406 billion dollars, according to the World Bank’s International Debt Report [4]. Countries eligible for the bank’s concessional IDA window paid a record 96.2 billion dollars, and rates on loans from official creditors doubled to more than 4% ✓ Established Fact [4]. Chinese loans were already priced above the concessional norm — AidData finds a typical Chinese development loan carries 4.2% interest and a maturity of under 10 years, against 1.1% and 28 years for OECD donors such as Germany, France and Japan [11]. When the global cost of money rose, the most expensive bilateral debt on the books was also the fastest to mature.

The question this report tests is therefore narrower and harder than the one usually asked. It is not whether China lends too much, or whether its projects are wise; the answer to both varies by country. It is whether the specific claim — that Beijing lends in order to capture strategic assets when borrowers default — is supported by the loan-by-loan record, and if not, what the record does support. The following sections trace the claim from its origin in a 2017 newspaper column, test it against its founding case in Sri Lanka, read the contracts, follow seven borrowers through distress and examine the rescue-lending machine that has quietly replaced the Belt and Road as China’s dominant form of overseas credit ◈ Strong Evidence [5][14].

The term was coined by Brahma Chellaney, a strategic analyst at the Centre for Policy Research in New Delhi, in a Project Syndicate column published on 23 January 2017 [5]. His argument was that Belt and Road lending to strategically located developing countries was designed to create dependency, and that debt would be converted into leverage over ports and other assets ⚖ Contested [5]. The framing was not built on a dataset. It was a strategic reading of a small number of visible projects, chiefly in South Asia, written from the vantage point of India — the country with the most direct interest in how China’s presence in the Indian Ocean was interpreted [7]. That origin does not make the argument wrong. It does explain why its evidence base was so thin at the moment it was adopted.

The adoption was extraordinarily fast. Deborah Brautigam, who directs the China Africa Research Initiative at Johns Hopkins, traced how the phrase moved from an Indian think tank through media, intelligence circles and Western governments, generating nearly 2 million search results within 12 months ✓ Established Fact [7]. In March 2018 the Center for Global Development published the first systematic debt-risk screen of the Belt and Road, finding that of 68 countries hosting projects, 23 were already at risk of debt distress and in eight of them — Pakistan, Laos, the Maldives and Mongolia among them — Belt and Road financing would significantly increase that risk [10]. A month later, a Center for Strategic and International Studies analysis described how China bought Hambantota [16].

The CGD study is worth reading closely because it is so often cited as evidence for a claim it did not make. Its authors warned about debt sustainability, and they were right: several of the eight countries have since required rescheduling or deferral [10][24]. But they did not argue that China intended to seize assets, and they recommended better lending practices rather than containment [10]. By October 2018 the distinction had collapsed. Vice-President Mike Pence, speaking at the Hudson Institute, accused China of debt diplomacy and presented Hambantota as its proof ✓ Established Fact [6].

The speech converted an analytical hypothesis into state doctrine. Pence said China offered hundreds of billions of dollars in infrastructure loans to governments from Asia to Africa to Europe and Latin America, that the terms of those loans were opaque at best, and that the benefits flowed overwhelmingly to Beijing [6]. The final sentence of his Hambantota passage was the one that travelled. Over the next five years the phrase would be used by successive American administrations, congressional committees and finance ministries across Asia and Africa [7]. Beijing’s reply was equally categorical. In August 2022 foreign ministry spokesperson Wang Wenbin called the Chinese debt trap a lie made up by the US and some other Western countries ✓ Established Fact [37].

Wang’s rebuttal came with numbers, and they deserve to be tested rather than dismissed. Citing World Bank data, he said China held less than 10% of the public external debt of low and lower-middle-income countries, against 40% held by commercial creditors and 34% by multilateral institutions [37]. As an aggregate, the claim is broadly consistent with independent data. As a description of the countries at the centre of the debate, it is misleading: in Laos China holds nearly half of sovereign external debt, and in Djibouti almost 50% ◈ Strong Evidence [24][35]. Averages across 100-plus countries dilute exactly the concentrations that matter.

Both sides of the argument thus inherited a flaw from the way it began. The trap thesis generalised from a single case whose facts had not been checked; the rebuttal generalised from aggregates that hide the cases in which concentration is extreme. What neither side had in 2018 was the evidence that would eventually settle most of the dispute: the contracts themselves, the rescue-lending data and the outcomes of real restructurings [12][14][26]. Between 2021 and 2025 all three became available. The meme was built before the data. The rest of this report reads the data.

The port was a Sri Lankan idea long before it was a Chinese loan. The concept of a deepwater harbour at Hambantota dates to the 1970s and a local parliamentarian, D. A. Rajapaksa; his son Mahinda, as minister responsible for ports, gazetted the project in 2001 and later made it a centrepiece of his presidency [9]. Sri Lanka first approached the United States and India for financing, and both declined ✓ Established Fact [8]. The construction contract went to China Harbour Engineering in 2007 — six years before the Belt and Road Initiative was announced [8]. The first phase was financed by a 307 million dollar China Eximbank loan at 6.3% interest, well above the 2% to 3% typical of multilateral development banks ✓ Established Fact [16].

The port then failed commercially, for reasons that had more to do with Sri Lankan governance than with Chinese strategy. Colombo handled 5.7 million twenty-foot equivalent units of container traffic in 2016, while Hambantota received only 175 cargo ships in the year before CSIS published its analysis [16]. By 2015 a new government had halted construction and opened renegotiations [16]. In 2017, under pressure to raise foreign currency, Sri Lanka leased 70% of the port to China Merchants Port Holdings for 99 years in exchange for 1.12 billion dollars, completing the handover in December 2017 [16]. This is the transaction that became the emblem of the debt trap.

The lease itself was a commercial negotiation, and a hard one. Chatham House researchers Lee Jones and Shahar Hameiri document that China Merchants Port paid 973.7 million dollars for its stake and committed a further 146.3 million, disputed the port’s valuation and demanded an adjacent 15,000-acre site for an industrial zone to support its port-park-city model [9]. The company had pursued the same model elsewhere: it had bought a 49% stake in a French port operator in 2013 and taken a 99-year lease on the Australian port of Newcastle [9]. As the researchers note, long leases are normal in the port sector, where investments take two decades to pay back; Australia has leased five ports for 98 to 99 years ✓ Established Fact [9].

The debt arithmetic is the strongest evidence against the trap reading. When Maithripala Sirisena took office in 2015, Sri Lanka owed more to Japan, the World Bank and the Asian Development Bank than to China, and only about 5% of the 4.5 billion dollars Sri Lanka paid in debt service in 2017 was attributable to Hambantota ✓ Established Fact [8]. Chinese loans were roughly 20% of foreign debt in 2022, while international sovereign bonds — held mainly by Western investment funds — accounted for about 36% [17]. Sri Lanka defaulted in 2022 because it had borrowed heavily in international capital markets and run down its reserves, not because of a single Chinese loan [9][17]. When restructuring came, China Eximbank agreed terms on about 4.2 billion dollars of debt in October 2023 rather than enforcing on assets [18].

The military claim has also not materialised in the form Pence predicted. Under the lease, port security is assigned to an oversight committee comprising the Sri Lankan navy and police, the ports authority and a government secretary, with the Chinese operator responsible only for internal security; in 2018 Sri Lanka decided to relocate its navy’s southern command to the port ◈ Strong Evidence [9]. Jones and Hameiri found no evidence of Chinese military activity at or near Hambantota after the lease began, while US and Indian naval vessels had made port visits [9]. Sri Lanka’s ambassador to Beijing put it flatly: China never asked, and Sri Lanka never offered [9].

Hambantota therefore teaches the opposite of what it was used to teach. The debt-trap chain — unsustainable loan, default, seizure, military base — broke at every link: the loan was not the cause of default, there was no seizure, and there is no base [8][9]. What remains is subtler. A Chinese state firm holds a 99-year commercial position in a strategically located port, and Sri Lanka has had to manage Indian and American pressure over what docks there ◈ Strong Evidence [19][20]. That is influence through ownership and proximity. It is real, and it deserves scrutiny. It is not a debt trap, and calling it one has made it harder to see.

The study, How China Lends, was produced by Anna Gelpern, Sebastian Horn, Scott Morris, Brad Parks and Christoph Trebesch for AidData, the Center for Global Development, the Kiel Institute and the Peterson Institute [12]. It compared 100 contracts between Chinese state lenders and government borrowers in Africa, Asia, Eastern Europe, Latin America and Oceania with the contracts of other bilateral, multilateral and commercial creditors [12]. Its conclusion was that Chinese contracts are muscular commercial instruments with unusual features: confidentiality that bars borrowers from revealing the terms or even the existence of the debt, collateral arrangements that give lenders a claim over revenue, and clauses that could let the lender influence debtors’ domestic and foreign policies ✓ Established Fact [12].

The numbers are specific. All China Development Bank contracts in the sample, and 43% of China Eximbank contracts, contain confidentiality clauses; every Eximbank contract signed after 2014 does [12]. Close to three-quarters contain what the authors call No Paris Club clauses — undertakings to exclude the debt from any collective restructuring or comparable treatment — including 81% of Eximbank and all CDB contracts ✓ Established Fact [12]. 98% contain cross-default clauses, and 29% use liens, escrow or special accounts, including six of eight CDB loans [12]. All CDB contracts in the sample list the termination of diplomatic relations with China as an event of default [12]. That last clause is the closest the documents come to explicit political leverage.

Where collateral is used, it is typically cash, not concrete. The 2025 How China Lends 2.0 dataset shows Chinese state lenders securing priority repayment by routing developing countries’ commodity export revenues into offshore bank accounts that the lender can draw on ✓ Established Fact [13]. Ecuador is the clearest example. Since 2010 Chinese policy banks have lent it more than 18 billion dollars, paired with long-term oil supply contracts to PetroChina and Unipec, so that repayment flowed through committed barrels regardless of market price [30]. By 2013 nearly 90% of Ecuador’s oil exports were committed under term contracts with Chinese buyers, and a 2022 audit estimated nearly 5 billion dollars in revenue lost to below-market pricing ◈ Strong Evidence [30].

Clauses that look like asset grabs often turn out, on inspection, to be standard sovereign-lending boilerplate or misreadings. Montenegro’s 944 million dollar China Eximbank loan for the Bar-Boljare motorway, signed in October 2014 at 2% over 20 years with a six-year grace period, contains an Article 8.1 waiver of sovereign immunity for arbitration in Beijing that excludes diplomatic and military property [33]. Commentators read this as a pledge of Montenegrin land; the clause does not say so, and immunity waivers are routine in commercial sovereign debt [33]. In Kenya, the claim that Mombasa Port secured the Standard Gauge Railway loans arose from the auditor-general wrongly labelling Kenya Ports Authority as a borrower; the released contracts show the Republic of Kenya is the borrower and the port was never pledged ✓ Established Fact [21].

Secrecy has consequences beyond the individual contract. AidData’s 2021 Banking on the Belt and Road study, covering 13,427 projects worth 843 billion dollars in 165 countries, found about 385 billion dollars of debt owed to China that did not appear in official statistics, because it was contracted by state-owned companies, banks and special-purpose vehicles rather than by central governments ◈ Strong Evidence [11]. 42 low and middle-income countries had debt exposure to China above 10% of GDP, and the average government was underreporting its actual and potential obligations to China by an amount equal to 5.8% of GDP [11]. Underreporting rose from 13 billion dollars a year before the Belt and Road to 40 billion during it [11].

Hidden debt is the mechanism by which the contracts produce leverage without seizure. A finance ministry that cannot disclose the full terms of its obligations cannot present a complete picture to the IMF, to bondholders or to its own parliament; a creditor that holds collateral accounts and a No Paris Club undertaking can decline to join a collective restructuring while others take losses [12][13]. Kenya’s Court of Appeal ruled in 2020 that the procurement of the China-funded railway contract was unlawful, and the contracts were released only after sustained litigation [22][21]. The trap, where it exists, is informational and legal. It is not written in the language of ports.

Zambia was the first test under the G20’s Common Framework. It defaulted in November 2020, the first African sovereign to do so during the pandemic, owing Chinese lenders somewhere between 3.3 billion dollars in official reports and 6.6 billion in independent research — about 30% of its external debt [27]. The trap thesis predicted that Beijing would use the default to extract concessions over copper or infrastructure. Instead China co-chaired the official creditor committee with France, and in June 2023 the committee agreed a treatment of 6.3 billion dollars of official debt ✓ Established Fact [26][27]. The process took two and a half years and was slowed by disputes over which Chinese lenders counted as official, but it ended in relief rather than foreclosure [27][28].

Laos is the hardest case for the critics of the trap thesis, and it should be treated as such. The Lowy Institute estimates Lao public and publicly guaranteed debt at about 112% of GDP in 2023, with China holding nearly half of sovereign external debt — roughly 5.1 of 10.5 billion dollars ✓ Established Fact [24]. Beijing has deferred about 2.5 billion dollars of payments, the kip has lost 50% of its value against the dollar since early 2022 and inflation has run in double digits [24]. The debt was driven by a 6 billion dollar China-Laos railway and by a hydropower build-out financed in large part by Chinese lenders that created far more generating capacity than the country could use or sell [24].

The decisive Lao transaction happened in 2020 and 2021. China Southern Power Grid bought 90% of a new transmission company, EDL-Transmission, for about 600 million dollars, and in March 2021 the venture signed a 25-year concession giving it control of the high-voltage grid — and, with it, the country’s electricity exports to its neighbours ✓ Established Fact [25]. At the time Laos had sovereign debt of about 12.6 billion dollars, around 65% of GDP, plus an estimated 8 billion dollars at the state utility [25]. This is not a foreclosure: the transfer was negotiated and paid for. But it is the closest thing in the entire record to what the trap thesis describes — a strategic national asset passing into Chinese state control in the middle of a Chinese-financed debt crisis.

Kenya shows the opposite trajectory: a borrower that used litigation and currency markets to improve its terms. Kenya borrowed about 5 billion dollars from China Eximbank for the Standard Gauge Railway from Mombasa towards Nairobi [23]. After courts forced disclosure of the contracts, it emerged that Mombasa Port had never been pledged [21][22]. In October 2025 the Treasury converted the dollar-denominated loans into yuan, cutting the interest rate from about 6.37% and saving roughly 215 million dollars a year ✓ Established Fact [23]. The conversion deepens Kenya’s exposure to Chinese currency, which is its own form of dependence, but it is a renegotiation on Kenyan initiative, not an enforcement.

Pakistan and Djibouti illustrate the rollover model. In June 2025 China rolled over 3.4 billion dollars of loans to Pakistan, including a 2.1 billion dollar deposit at the State Bank of Pakistan and refinancing of a 1.3 billion dollar commercial loan that had already been repaid ✓ Established Fact [36]. In Djibouti, where China holds almost 50% of external debt and opened its only African military base in 2017, China Eximbank agreed in October 2023 to suspend payments on the 2013 railway loan for 2024 to 2027 [34][35]. The Chinese base predates the moratorium by six years [35]. The base came from diplomacy and rent, not from default.

Malaysia and Ecuador complete the picture. In April 2019 Malaysia’s new government renegotiated the East Coast Rail Link with China Communications Construction Company, cutting its construction cost from RM65.5 billion to RM44 billion by shortening and realigning the route ✓ Established Fact [32]. Ecuador’s Coca Codo Sinclair hydropower plant, financed by a 1.7 billion dollar China Eximbank loan, had about 17,499 documented cracks by July 2022, yet supplies more than 30% of the country’s electricity [31]. Across the seven cases the pattern is consistent: bad projects, expensive money, long negotiations — and Chinese creditors choosing deferral over enforcement ◈ Strong Evidence [28].

The single most important finding of the past five years is not in a contract but in a balance sheet. Sebastian Horn, Brad Parks, Carmen Reinhart and Christoph Trebesch documented that China has become an international lender of last resort for its own borrowers, using rollovers, balance-of-payments loans and central-bank swap lines [14]. The People’s Bank of China’s swap network alone provided more than 170 billion dollars of liquidity to countries in crisis [14]. The share of China’s overseas lending portfolio going to countries in debt distress rose from less than 5% in 2010 to 60% in 2022, and nearly 80% of the rescue lending was issued between 2016 and 2021 ✓ Established Fact [15].

The rescue loans are expensive and opaque. They carry an average interest rate of about 5%, against 2% for IMF rescue lending, and are almost exclusively targeted at Belt and Road debtors [14][15]. Reinhart’s reading of the motive is direct: Beijing is ultimately trying to rescue its own banks, and that is why it has gotten into the risky business of international bailout lending [15]. A creditor that lends new money to keep old loans current is not setting a trap; it is avoiding the recognition of losses. But the effect on the borrower can be similar, because each rollover deepens the dependence on a single lender whose terms cannot be seen.

The shift has changed what Chinese credit is for. AidData’s 2025 figures show China still lent about 140 billion dollars in 2023, making it the world’s largest official creditor, but official development assistance fell to 1.9 billion dollars, the lowest in 20 years ✓ Established Fact [2]. Lending has moved towards high-income and upper-middle-income economies, towards liquidity facilities and towards sectors such as critical minerals and semiconductors [1][2]. Chinese lending to Africa illustrates the collapse of the old model: 1,306 loans worth 182.28 billion dollars between 2000 and 2023, but only 4.61 billion dollars in 13 new commitments in 2023, against more than 10 billion a year at the Belt and Road’s height [29].

The leverage this creates is real but different from the one the trap thesis described. A borrower that needs an annual rollover from Beijing to meet its IMF programme targets cannot afford a diplomatic rupture, and every CDB contract in the 2021 sample lists termination of diplomatic relations as an event of default ◈ Strong Evidence [12][36]. That is influence over foreign policy exercised through the calendar of refinancing — quieter than a naval base and far more common. Parks observes that China’s strictly bilateral approach has made it harder to coordinate the activities of all major emergency lenders [15]. The IMF, the Paris Club and the bondholders now negotiate around a creditor whose exposure they cannot fully see.

The risk table above ranks the mechanisms by what the evidence supports, not by how often they appear in speeches. Opacity ranks first because it corrupts everything downstream: an IMF programme designed around reported debt will be wrong if 5.8% of GDP is missing from the ledger ◈ Strong Evidence [11]. Seniority clauses rank second because they determine who absorbs losses when relief comes [12]. Asset transfer ranks last not because it is impossible — Laos shows it is not — but because it is the rarest of the five mechanisms in the documented record [24][25]. The popular debate has the order almost exactly inverted.

The global rate environment amplifies all five. Developing countries paid a record 406 billion dollars in interest alone in 2023, and official lenders’ rates doubled to more than 4% [4]. Chinese debt, shorter and dearer than Western concessional lending, matures into that environment first [11]. That is why the Lowy Institute’s peak-repayment window of the mid-2020s matters more than any single port: the pressure on budgets for health, education and climate adaptation is being exerted now, by scheduled repayments, in 75 of the poorest countries ✓ Established Fact [3].

The strongest critique of the trap thesis is structural. Jones and Hameiri argue that China’s development-finance system is too fragmented and poorly coordinated to pursue detailed strategic objectives, that economic factors drive most projects and that recipient governments and their domestic interests determine what gets built ◈ Strong Evidence [9]. In Sri Lanka and Malaysia — the two most widely cited victims — the controversial projects were initiated by the recipient governments for their own purposes [9]. Brautigam and Meg Rithmire add that after years of examining Chinese loan documents, they found no evidence that China seizes assets when borrowers fail to pay [8].

The strongest defence of a modified thesis comes from the cases the critics treat least fully. The Lowy Institute’s Laos team accepts domestic responsibility but concludes that China lent massively to weak institutions and created an energy-sector overcapacity that the country cannot service ⚖ Contested [24]. The How China Lends authors show that even if Chinese contract terms were unenforceable in court, the mix of confidentiality, seniority and policy clauses could limit a debtor’s crisis-management options and complicate renegotiation [12]. Neither argument requires a master plan. Both describe a system whose incentives produce leverage whether or not anyone intended it.

The second disagreement is about responsibility for the crises themselves. Sri Lanka’s default involved international sovereign bonds that were about 36% of public foreign debt, against roughly 20% owed to China ◈ Strong Evidence [17]. Zambia’s default involved Eurobonds and commercial creditors alongside Chinese lenders [27]. Beijing’s foreign ministry uses exactly these figures to argue that Western commercial creditors, not China, hold the dominant claims [37]. The counter-argument is that aggregate shares understate China’s role in the specific countries — Laos, Djibouti, Zambia — where its lending was concentrated, and that expensive short-maturity bilateral debt shaped the timing of distress [24][35][11].

The third disagreement — whether China obstructs collective relief — has shifted most with new evidence. The contracts are designed to keep Chinese debt outside Paris Club treatments, and the Zambia process was slowed by disputes over which Chinese lenders counted as official [12][27]. Yet China co-chaired the Zambian creditor committee and signed the 2023 treatment, and the Oxford Business Law Blog reads its role there as a pragmatic interest in stabilising debtor economies ⚖ Contested [27][26]. In Ethiopia, the official creditor committee signed a memorandum in July 2025, while it is private Eurobond holders who rejected an 18% principal reduction as failing comparability of treatment [38].

Set side by side, the two columns are less contradictory than their labels suggest. The myth column is correct about seizure, about who initiated the showcase projects and about the dominant response to distress [8][9][28]. The trap column is correct about concentration, about contractual design and about hidden debt [3][11][12]. The synthesis is that China behaves less like a strategist laying traps than like a large, state-directed creditor protecting itself — and that a creditor of that size, protecting itself through secrecy and seniority, produces political leverage as a by-product ◈ Strong Evidence [14].

What remains genuinely unresolved is intent at the margin. The diplomatic-relations default clause, the Lao grid concession and the Hambantota survey-ship visit are all consistent with opportunism as well as design [12][25][19]. The data cannot read minds. What it can show is that the outcomes feared in 2018 — foreclosed ports, naval bases acquired through default — have not occurred in the seven years since, while the outcomes few people predicted — mass rollovers, negative net flows and 240 billion dollars of rescue lending — have ✓ Established Fact [3][14].

The first conclusion is that the headline claim is not supported. The founding case, Hambantota, was a lease without debt relief on a project Sri Lanka proposed, whose proceeds went to reserves and whose security remains in Sri Lankan naval hands [8][9]. The most repeated African claim, Mombasa Port, was a clerical misreading [21]. The most repeated European claim, Montenegrin land, rests on a standard immunity waiver [33]. Across 2.2 trillion dollars of lending, researchers have not documented a single strategic asset foreclosed for non-payment ◈ Strong Evidence [7][8].

The second conclusion is that the thesis pointed at a real problem and misnamed it. China is now the dominant bilateral creditor to the developing world, in 2025 it collects a record 35 billion dollars in debt service, and its contracts are designed to make it senior, secret and separate from collective restructuring [3][12]. Those three features — seniority, secrecy and separateness — are what give a creditor leverage over a sovereign in distress. They are also, in milder form, features of commercial bond markets. The difference is scale, state direction and the addition of a political default clause [12].

The third conclusion is that the most important change in Chinese lending since 2016 is the one least discussed. The Belt and Road has given way to a rescue system: 240 billion dollars in bailouts, 170 billion in swap lines and 60% of lending directed to countries in distress by 2022 ✓ Established Fact [14][15]. At the same time the portfolio has migrated towards high-income economies, with the United States the largest single recipient [2]. The developing world is now mostly paying China back rather than borrowing from it, and the negative net flow of 34 billion dollars in 2024 is the arithmetic of that transition [3].

The fourth conclusion concerns responsibility. The evidence does not allow either side the simple story. Sri Lankan, Malaysian and Lao elites chose the projects that failed, often for political reasons, and Western bondholders were central to the defaults in Sri Lanka and Zambia [9][17][27]. But Chinese lenders financed those choices at higher rates and shorter maturities than concessional alternatives, kept the terms secret and built in protections that pushed the cost of failure onto other creditors and onto the borrowers’ citizens [11][12]. Shared responsibility is not equal responsibility; it varies by case, and Laos sits at the far end of the range [24].

The fifth conclusion is about remedies. If the problem is opacity, the remedy is disclosure: the How China Lends contract repository and the hidden-debt estimates exist only because researchers assembled them from leaked, litigated and published documents [12][11]. Kenya obtained its contracts through the courts [22]. A norm under which sovereign borrowers publish every loan contract, and lenders accept that confidentiality clauses are void in public-debt contracts, would neutralise the most dangerous feature of the current system without requiring anyone to prove intent ◈ Strong Evidence [12].

What the evidence will not support is either of the two stories that dominated the past decade. The trap story — a grand strategy of engineered default and seized ports — is contradicted by the documents, the restructurings and the absence of a single foreclosure [8][9][28]. The myth story — China as a normal lender maligned by rivals — is contradicted by the confidentiality clauses, the collateral accounts, the political default trigger and the 385 billion dollars of hidden debt [12][11]. The accurate description is less dramatic and more durable: the largest official creditor in the world has built its lending to be repaid first, and in 75 of the poorest countries, repayment is now what the relationship consists of ✓ Established Fact [3].