The Chinese Infrastructure Binge and the Road to Default

The asset side of the bargain

Article 14 examined the GTP-era growth story from the macroeconomic ledger: the inflation, the parallel premium, the accumulating external debt. This article examines the same period from the asset side — the specific infrastructure that the borrowing built, the terms on which it was financed, and the question of whether the assets generated the cash flows necessary to service the debt that funded them. The answer, by 2023, was that they had not, and the resulting Eurobond default — Ethiopia’s first sovereign default in modern history, and the then-largest in African history — was the visible breaking point.

The argument is not that Ethiopia should not have built infrastructure during the GTP years. The infrastructure was, in important cases, genuinely needed. The argument is narrower and more specific: that the financing terms were structured around assumptions — about future growth rates, future export earnings, future exchange-rate stability — that turned out to be wrong, and that the discipline that should have constrained the borrowing was, by design, absent. The Chinese commercial loans and the Eurobond issuance were not the principal cause of Ethiopia’s monetary disorder. They were the mechanism through which the existing monetary disorder produced a sovereign-debt crisis on a timeline.

The China relationship

Chinese lending to Ethiopia, of all forms, has totalled approximately $14 billion since 2000, making Ethiopia the second-largest African recipient of Chinese finance after Angola.¹ The lending has come through multiple channels: China Eximbank concessional and commercial loans, China Development Bank facilities, Sinosure-backed supplier credits, and direct financing by Chinese state-owned enterprises operating as project sponsors. The terms have varied widely — from concessional rates with long grace periods on early projects, to commercial rates indexed to LIBOR-plus-margin on later ones.

The three largest single projects on the Ethiopia side are worth examining individually because their economics illustrate the broader pattern.

The Addis Ababa–Djibouti Railway was financed by China Eximbank through a $2.49 billion buyer’s credit loan signed in May 2013, with a 15-year maturity, a 6-year grace period, and an interest rate of LIBOR + 3 percent margin.² The project was constructed by China Railway Engineering Corporation and China Civil Engineering Construction Corporation, completed in 2016, and inaugurated as the first cross-border electrified railway in Africa since the 1970s.³ The economic case was that the railway would slash cargo trips from weeks to hours, capture a major share of Ethiopia’s foreign trade (more than 90 percent of which passes through Djibouti), and catalyse industrial development along its route.

The economic reality, as documented by the Overseas Development Institute and others, has been more complicated. The railway has faced persistent operational problems: electricity-supply instability that has interrupted services, flooding-related damage, last-mile connectivity issues that have limited the line’s catalytic effect on industrial parks, and revenue from passenger and freight operations far below the original projections.⁴ Sinosure, the Chinese export-credit insurer, reportedly wrote off approximately $1 billion in losses on the railway because of its poor commercial performance.⁵ The Djibouti government formally requested restructuring of its share of the loan in late 2017, and a restructuring was eventually agreed in 2020 with extended maturity and reduced effective interest rates.⁶ Whether the Ethiopian share has been similarly restructured has not been publicly disclosed in equivalent detail.

The Grand Ethiopian Renaissance Dam was, by design, principally financed by Ethiopian sources rather than by Chinese loans. The $5 billion total project cost was raised largely through Ethiopian government bonds, employee contributions (effectively a payroll deduction), and diaspora bonds.⁷ But the transmission infrastructure that connected the dam to the grid — roughly $1.5 billion in 2013 — was financed by China, with an additional $1.2 billion loan in the same year and a further $1.8 billion offered in 2019 after Abiy Ahmed’s Beijing visit.⁸ The dam was finally inaugurated in September 2025, with a generating capacity of 5,150 MW — the largest hydroelectric facility in Africa by capacity.⁹

Industrial parks, telecoms infrastructure, the new light-rail in Addis, the airport expansion, the Awash–Hara Gebeya railway (Turkish-financed), and a number of other projects collectively absorbed several billion dollars of external financing across the decade. Each had its own justifications, its own contractors, its own financing terms. The cumulative effect was a substantial expansion of Ethiopia’s external debt stock — from roughly $9 billion in 2010 to over $30 billion by 2018 and approximately $60 billion by 2023.¹⁰

What the borrowing did not produce

The bet underlying the borrowing was, as article 14 argued, that the infrastructure would generate the growth and the exports needed to service the debt. The bet has not paid off on the timeline assumed.

Ethiopian goods exports, which the GTP-era plans projected would more than double over the decade, grew — but at substantially slower rates than projected and from a smaller base than the imports the country was simultaneously committed to fund. By the late 2010s, Ethiopia’s current-account deficit was running at 6–8 percent of GDP, with import demand chronically exceeding export earnings, requiring continuous external financing to bridge the gap.¹¹ The export base remained heavily concentrated in coffee, sesame, gold, and a narrow set of other commodities; the manufactured exports the industrial parks were supposed to generate emerged slowly and at smaller volumes than projected.

The infrastructure-driven growth that should have, in the developmental-state framework, broadened the productive base and generated diversified export earnings was, in important respects, consumption-shifting rather than production-enhancing. The railway moved existing trade more efficiently but did not, in the timeframe assumed, generate the new export industries the catalytic effect was supposed to produce. The dam expanded power-generation capacity but did not, before its 2025 inauguration, contribute meaningful new export earnings (electricity exports to neighbouring countries grew, but at small volumes relative to debt-service requirements). The industrial parks attracted some foreign investors, particularly in textiles and apparel, but the export earnings these generated were smaller than projected and were partially offset by the import-intensity of the same industries.¹²

The mathematics of the situation became, by 2020, increasingly difficult to ignore. External debt service was rising as the grace periods on early loans expired and amortisation began. Foreign-exchange earnings were growing more slowly than projected. The cushion between earnings and obligations was narrowing. The 2020 onset of the Tigray war and the simultaneous COVID-19 pandemic — examined in article 20 — compressed the cushion further, with AGOA suspension, donor freezes, and tourism collapse all reducing inflows that the debt service plan had been counting on.

The November 2023 default

The 24 November 2023 missed coupon payment on the December 2024 Eurobond — a $33 million interest payment, against the bond’s $1 billion face value at a 6.625 percent coupon — was Ethiopia’s first sovereign default in modern history.¹³ It was, at the time, the largest sovereign default in African history. It triggered a series of consequences that the country is, in mid-2026, still managing.

The Ethiopian government’s framing of the default, articulated by State Minister of Finance Eyob Tekalign in late 2023, was that it was not a sign of insolvency but a strategic alignment of debt servicing across bondholders and official creditors under the G20 Common Framework Agreement.¹⁴ The argument was that the Common Framework required burden-sharing across all creditor classes — bilateral, multilateral, and commercial — and that paying the Eurobond while restructuring with bilateral creditors would have violated the framework’s comparability-of-treatment principle. The default was, on this account, technical rather than substantive.

The argument has some merit but does not fully capture the picture. The Common Framework comparability principle does indeed require burden-sharing; but the principle does not require the default to occur before negotiations. Ethiopia could have continued paying the coupon while negotiating restructuring terms; the choice not to do so was a choice that traded a near-term cost (the political and reputational consequences of default) against a near-term benefit (the negotiating leverage the default created). Whether the trade was worth it is debatable. The default did, in fact, push the bondholders toward eventual negotiation; whether it pushed them faster or further than continued payment would have done is unprovable.

The largest holders of the Eurobond were American Beacon Frontier Markets Fund ($125 million of the $1 billion total), Templeton Emerging Markets Bond Fund ($65 million), Pictet ($51 million), and JP Morgan’s Emerging Markets Bond Fund ($31 million), with the remainder distributed across dozens of other mutual funds and institutional investors.¹⁵ These are not predatory creditors; they are pension funds and emerging-market bond funds whose portfolios serve retail investors and institutional savers in the United States and Europe. The default cost their beneficiaries money. It also damaged Ethiopia’s reputation in the markets these funds participate in.

The Common Framework process

The G20 Common Framework for Debt Treatments was created in 2020 as a successor to the Heavily Indebted Poor Countries (HIPC) initiative, intended to provide a structured mechanism for sovereign-debt restructuring that included China and other major non-Paris-Club creditors as well as the traditional Paris Club and commercial creditors.¹⁶ Ethiopia formally applied for treatment under the Common Framework in early 2021. The process has been long.

The interim debt service suspension was agreed with the Official Creditor Committee in November 2023, suspending payments on bilateral and private credits maturing between January 2023 and December 2024.¹⁷ The Agreement in Principle with the OCC was reached in March 2025, with China — the largest bilateral creditor, holding approximately 25 percent of Ethiopia’s external debt — leading the consortium of bilateral creditors.¹⁸ The Memorandum of Understanding formalising the agreement was signed in July 2025, providing approximately $3.5 billion in cashflow relief.¹⁹

Bondholder negotiations have been more difficult. The ad hoc committee of bondholders refused the initial Ethiopian restructuring proposals through 2024 and 2025; talks broke down in October 2025 without agreement.²⁰ A preliminary agreement was announced in January 2026, scrapped within weeks when the official creditors objected to its terms, and a renewed agreement was announced on 29 June 2026, on substantially different terms: a new bond worth $880 million (12 percent haircut from the $1 billion original), 6.15 percent interest rate, repaid in annual installments between July 2026 and July 2029, with the three missed coupon payments ($99.375 million total) paid alongside, plus a consent fee.²¹ Whether this agreement holds remains to be seen; the bondholders have, in past iterations, agreed and then withdrawn.

The Fitch Ratings designation of Ethiopia as “Restricted Default” in October 2025 reflected the continued default on commercial obligations even as bilateral restructuring progressed.²² The rating consequences — restricted access to international capital markets, higher implied costs of future borrowing — will persist for some period after the bond restructuring is finalised.

The lessons

The Eurobond default and the Common Framework process have produced several lessons that the series’s later articles will draw on.

First, the bet did not pay off on the timeline assumed. The infrastructure that was built has not, by 2026, generated the export earnings or the growth dividends necessary to service the debt that funded it. The bet may yet pay off over a longer horizon — the GERD’s full export potential will take years to materialise; the industrial parks may yet attract major investment; the railway may yet generate the catalytic effects originally projected. But the timeline mattered, and the timeline did not work.

Second, the financing terms were structured around macroeconomic conditions that did not hold. The LIBOR + 3 percent rate on the Addis-Djibouti railway loan was reasonable when LIBOR was 0.5 percent and the birr was at 17 to the dollar. With LIBOR at higher rates and the birr at 157 to the dollar, the same loan represents a substantially heavier burden in birr terms than the original projections assumed. Currency risk on dollar-denominated infrastructure debt was, by 2024, the single largest contributor to the country’s debt-sustainability problem.

Third, the lack of an open and disciplined domestic capital market made commercial external borrowing the only available large-scale financing option. A country with a developed bond market in its own currency could have financed its infrastructure with birr-denominated bonds, avoiding the currency-risk problem. Ethiopia did not have that option, because the conditions for a deep domestic bond market — credible monetary policy, positive real interest rates, an independent central bank — were precisely the conditions the developmental-state model had foreclosed. The dollar-denominated borrowing was, in this sense, downstream of the same architectural choices article 12 examined.

Fourth, China is now a major sovereign creditor with its own institutional preferences. The Common Framework process has been, in significant part, a negotiation about how China’s bilateral lending fits into the established creditor architecture. Ethiopia’s experience is one of several test cases — Zambia, Ghana, and Chad are the others — in which the world is collectively learning how to restructure debt that includes large Chinese components alongside traditional Paris Club and commercial elements. The early evidence is that the process is slow, that no Common Framework restructuring has yet been fully completed, and that the framework has delivered limited aggregate relief — by some measures, only about 7 percent of the combined debt stock of the high-risk countries that have entered it.²³ The lessons for future African sovereign borrowing are still being learned.

The next article steps back to examine a pattern that has run across the entire history this series has tracked: how every Ethiopian regime has paid for war with the printing press.

---

Footnotes

1. “Stalled Dreams,” The Wire China, citing $14 billion in Chinese lending to Ethiopia since 2000. ↩

2. AidData Project 70086, China Eximbank $1.28 billion buyer’s credit loan for the Sebeta-Adama-Meiso section of the Addis Ababa-Djibouti Railway, https://china.aiddata.org/projects/70086/. ↩

3. Boston University Global Development Policy Center, “Financing Ethiopia’s Railways with China and Turkey,” 2021, https://www.bu.edu/gdp/2021/02/08/financing-ethiopias-railways-with-china-and-turkey/. ↩

4. AidData Project 70086, citing the ODI 2021 report on the railway’s operational challenges. ↩

5. AidData Project 70086, citing the Sinosure write-off of approximately $1 billion. ↩

6. AidData Project 46183, China Eximbank loan for the Ali Sabieh to Nagad section of the Addis-Djibouti Railway, restructuring history. ↩

7. Mongabay, “Ethiopia’s Renaissance mega-dam fuels energy hopes and regional anxiety,” February 2026, https://news.mongabay.com/2026/02/ethiopias-renaissance-mega-dam-fuels-energy-hopes-and-regional-anxiety/. ↩

8. Disruption Banking, “Why is China financing the Grand Ethiopian Renaissance Dam (GERD)?”, September 2023, https://www.disruptionbanking.com/2023/09/04/why-is-china-financing-the-grand-ethiopian-renaissance-dam-gerd/. ↩

9. Mongabay, “Ethiopia’s Renaissance mega-dam,” February 2026: “With a length of 1,780 meters (5,840 feet) and a capacity of 5,150 megawatts (MW), it is the largest hydroelectric dam in Africa by capacity.” ↩

10. World Bank, Ethiopia International Debt Statistics, 2010, 2018, and 2023 editions; “Tigray war,” Wikipedia, citing the $60.6 billion (3.3 trillion birr) figure as of 2023. ↩

11. International Monetary Fund, “Ethiopia: 2018 Article IV Consultation,” IMF Country Report 18/354 (December 2018). ↩

12. Cornelia Staritz, Hans-Otto Sano, and Lindsay Whitfield, “African Industrial Policy Revisited: Lessons from Ethiopia, Mozambique, and Senegal,” WIDER Working Paper, 2022. ↩

13. AddisFortune, “Ethiopia’s Eurobond Default: Now What?”, December 2023; CNBC Africa, “Key events in Ethiopia’s journey towards debt restructuring,” 2026. ↩

14. Eyob Tekalign, quoted in AddisFortune, “Ethiopia’s Eurobond Default: Now What?” ↩

15. UNDP, “From Debt to Development: Working Paper Series Ethiopia No. 3,” 2023, https://www.undp.org/sites/g/files/zskgke326/files/2023-04/UNDP%20-%20Shock%20Document%20-%20Working%20Paper%20Series%203%20-%20Final%20April%20132023.pdf. ↩

16. Addis Standard, “Five Years in Limbo: Ethiopia’s debt restructuring stalemate, IMF-backed G20 Common Framework failure,” December 2025. ↩

17. CNBC Africa, “Key events in Ethiopia’s journey towards debt restructuring.” ↩

18. CNBC Africa, “Key events,” and IMF press release on the fourth ECF review. ↩

19. Addis Standard, “Five Years in Limbo,” citing the July 2025 MoU with the OCC providing approximately $3.5 billion in cashflow relief. ↩

20. The EastAfrican, “Ethiopia bondholders’ $1bn debt restructuring talks collapse,” October 2025. ↩

21. Ecofin Agency, “Ethiopia and Bondholders Reach New Preliminary Deal on $1 Billion Eurobond Restructuring,” June 2026. ↩

22. Addis Standard, “Five Years in Limbo,” citing the October 2025 Fitch RD rating. ↩

23. Addis Standard, “Five Years in Limbo,” citing the ONE Campaign October 2025 analysis on aggregate Common Framework relief delivered to date. ↩