A G20 debt restructuring framework created in 2020 has still produced no final deal in the case Ethiopia filed in 2021, more than five years later

What is happening?

The G20 and the Paris Club created the Common Framework for Debt Treatments beyond the DSSI in November 2020. It is a coordination mechanism for restructuring the debt of the poorest sovereign borrowers, open to the 73 countries that were eligible for the Debt Service Suspension Initiative, and it works by bringing official bilateral creditors to a common position and then asking private creditors to deliver comparable treatment.

Its predecessor postponed payments rather than reducing them. The Debt Service Suspension Initiative ran from May 2020 to December 2021, 48 of the 73 eligible countries took part, and about 12.9 billion dollars of debt service was suspended.

Since November 2020, four countries have entered the Common Framework — Chad in January 2021, Ethiopia and Zambia in February 2021, and Ghana in January 2023.

One of those four cases had still not closed as of 2026-08-20. Ethiopia applied in February 2021 and reached a formal agreement with official creditors in July 2025, about 53 months later. Negotiations with holders of the single outstanding eurobond broke down on 2026-05-27; on 2026-06-01 the bondholder committee said publicly that it was considering litigation in English courts, and no source opened here reports that a case was filed; a new preliminary agreement carrying a 12 percent reduction of principal and a New Money Warrant was reported on 2026-06-29. That agreement was still at the stage of agreement in principle on 2026-08-20, with the exchange offer described only as expected in the coming months. Counted from the application, that is 66 months with no completed treatment.

The three other cases ended differently from each other. Chad reached an agreement in November 2022 that rescheduled maturities falling due in 2024 and did not provide immediate debt reduction. Zambia completed a bond exchange on 2024-06-12, about 40 months after applying, after missing an implementation deadline that the parties had set for themselves. Ghana reached an agreement in principle with official creditors on 2024-01-12, about 12 months after applying on that track alone, and signed a memorandum of understanding on 2024-06-11.

Whose problem is this?

The body that would have to move faster is the same body that wrote in its own annual report that the framework has to move faster. No participant can be compelled by another, and the interval between application and completion falls on people who are not at the table.

Where does this problem end?

The boundary here is the machinery of restructuring rather than the reasons debt accumulated in the first place. Two questions that sit immediately next to this one are deliberately left unanswered — how much debt ought to be written off, and who ought to bear the loss. This document measures how long the existing procedure takes and what it delivers, not whether the procedure aims at the right target.

What is the state now, and what should it be?

Now

Needs a new measurementthe target state: no source opened here gives a target duration for a Common Framework case, a deadline for any stage of one, or a target level of debt reduction. The Paris Club recorded in its 2025 annual report, as reported on 2026-06-24, that the framework has to deliver faster and bring every creditor in on a comparable burden, and it pointed to the confusion created by roughly 100 development finance institutions each asserting preferred creditor status. That report names a direction and no number. Nothing opened here states how long a finished case is supposed to take or what a finished case is supposed to deliver.

How big is it?

About 213 million people. That is the sum of the 2025 population estimates for the four countries that have entered the framework since November 2020 — Ethiopia at about 135.5 million, Ghana at about 35.1 million, Zambia at about 21.9 million and Chad at about 21.0 million.

This count measures people living in countries that entered the mechanism, not people who were harmed by the delay. It also mixes finished and unfinished cases, because two of the four were completed on the tracks reported here and one was not.

The circle one step wider cannot be drawn from the sources opened. All 73 eligible countries could in principle use the framework, and the World Bank reported in December 2025 that the 78 most vulnerable IDA-eligible countries carried 1.2 trillion dollars of external debt, but no source opened here gives the combined population of either group.

The widest frame belongs to a different question and is recorded here only as context. The United Nations reported in July 2023 that 3.3 billion people live in countries that spend more on debt interest than on health or education, and a secondary summary of the 2025 update puts global public debt at 102 trillion dollars for 2024. Neither figure is limited to countries that used this mechanism, and neither is used as a measure of it.

Under what conditions does it arise?

1. Participation is voluntary and comparability is not enforceable. The framework asks private creditors to deliver treatment comparable to what official bilateral creditors agree, but no source opened here identifies a legal instrument that compels it. Nothing in the design gives any participant the power to require another creditor to move first.

2. The creditor base is fragmented and seniority is contested. The Paris Club pointed in its 2025 annual report to about 100 development finance institutions each asserting preferred creditor status. Where seniority is unsettled, no creditor can price what it is being asked to give up.

3. The two tracks run in sequence rather than together. Official creditors settle first and private creditors afterwards, and the second stage begins its own negotiation from the start. Zambia reached its official creditor agreement in June 2023 and completed the bond exchange in June 2024. Ethiopia reached its official creditor agreement in July 2025 and had no completed exchange on 2026-08-20.

4. Leverage can sit with a single creditor. In the first case treated, about one third of external debt of roughly 3 billion dollars was in oil-backed lending owed to a single commodity trading company. A framework built to coordinate many creditors of similar size behaves differently when one claim is that large, and no rule opened here addresses that concentration.

5. No stage carries a deadline. In the Zambia case the parties set an implementation date of 2023-12-31 for themselves and passed it, and a revised agreement followed on 2024-03-25. A date that the parties set can be moved by the parties, and no external date exists to move against.

What has been tried?

Two directions were tried in sequence — suspend payments, then build a standing mechanism to restructure them. The first ended on schedule and reduced nothing by design. The second has no schedule.

What was found?

Why is it still unsolved?

Coordination failure — every participant is better off if the others move first, and nothing in the design decides who moves.

The framework was built to solve exactly this. Before it existed, a debtor negotiated with each creditor group separately and any concession made to one could be undercut by another. The framework brings official bilateral creditors to a single position and then asks private creditors for comparable treatment. What it never acquired is an instrument that makes the second half happen. Comparability is a request, and a request has no date attached.

The second part is that waiting is cheap for a creditor and expensive for a debtor. A creditor holding an unresolved claim keeps the claim. A country in an unfinished case keeps servicing what it can, stays outside normal market access, and waits. The costs of delay are therefore not shared by the parties that control the delay, which is the structural reason delay is not expensive to the side that can end it.

The third part is that the fragmentation the framework was meant to overcome has grown inside it. When roughly 100 development finance institutions each assert preferred creditor status, the size of the pool being divided is itself in dispute, and no creditor can judge whether a proposed treatment is comparable to anyone else.

A mechanism that nobody is obliged to finish looks the same whether it is working slowly or not working at all. There is no announcement when a case stalls. The application stands, the committees exist, statements are issued, and the difference between a case in its fourth year and a case that has quietly stopped is not visible from outside. The mechanism that assessed itself in 2025 asked for speed and set no date, which leaves the pattern exactly where it was.

What observation would mean it is solved?

Candidates — (a) a Common Framework case moves from application to a completed exchange within a published and repeated interval (b) the share of eligible countries that use the mechanism rises, rather than most distressed borrowers restructuring outside it (c) the net transfer on the debt of the poorest countries turns positive, or at least stops setting records in the other direction.

(a) alone measures paperwork. A finished document is not the same as money that stops leaving the country. The first case treated closed with a rescheduling and no immediate reduction, which counted as a completed treatment and moved no debt stock. Speed without depth would satisfy this observation and change nothing for a budget.

(b) alone can move for the wrong reason. Use of the mechanism could rise because it became faster, or because more countries fell into distress. Four applications out of 73 eligible countries is a low number and no source opened here establishes why the other 69 have not applied, so a rise cannot be read as an improvement without knowing what the non-use meant.

(c) alone is not attributable. The net transfer moves with interest rates, commodity prices and new lending far more than with any single restructuring mechanism. A better figure could arrive in a year in which no case closed at all. The three have to be read together, and (a) has to be read with the size of the reduction attached, not just the date of signature.

What is it connected to?

Fills with researchthe design of IMF programme conditionality that runs alongside these cases, the treatment of restructuring by credit rating agencies and any resulting reluctance to apply, collateralised and resource-backed lending as a category, the relationship between debt service and climate finance for the same countries, and proposals for a statutory sovereign debt workout mechanism. Relation type and evidence grade were not confirmed in this round.

What these sources do not say

• Why 69 of the 73 eligible countries have not applied. No source opened here analyses it. Whether those countries are not in distress, or expect an application to close off market access, is not answered — the possibility of a penalty for applying is suggested in commentary, and no opened source attaches evidence to it.
• What separates the completed cases from the open ones. Sources point to different factors, including the composition of the creditor base, the conduct of individual creditors and commodity prices. No opened source reconciles them into a single account.
• Any defence of the voluntary design. Every opened source that assesses the framework calls for it to be faster. None sets out why the current design is deliberate, and no public statement from the G20 or the Paris Club defending voluntary participation was located.
• The primary analysis behind the 7 percent figure. That number reaches this document through a news report that cites it. The analysis page itself returned HTTP 403 to automated requests, so neither its method nor its denominator was read here.
• Whether the open case will close, and on what terms. No source opened here states when the exchange offer will be launched, what the non-financial conditions still outstanding are, or whether the 12 percent reduction of principal will survive into a final agreement.
• The combined population of the eligible group. Neither the 73 DSSI-eligible countries nor the 78 IDA-eligible countries has a stated combined population in any source opened here, so the wider circle in the population estimate could not be drawn.
• What the treatments delivered per country. No opened source gives the amount of debt actually reduced in each of the four cases, which is the figure that would say whether a completed case is worth the years it takes.

See the evidence

No primary document of any debt treatment was read here. The World Bank questions and answers page, two World Bank statements, the World Bank press release of 2025-12-03, the Ghana ministry announcement and the four population dashboard pages were opened directly and carry the eligibility count, the suspension total, the population components and the aggregate debt figures. Everything about the four cases comes from agency reporting, professional commentary and one civil society research group; no agreement text, no creditor committee minute and no Paris Club report was opened. Three rows could not be opened at all and are left blank with the reason attached rather than filled with a substitute address. Where opened sources overlap they agree on dates: the 2024-03-25 bondholder agreement appears in two, and the stalled status of the longest open case appears in two from May and June 2026. Where they do not overlap the value is graded down in the findings table rather than presented as settled — the count of four applicants, the July 2025 official creditor agreement and the 7 percent share all rest on single or secondary sources, and the 7 percent figure in particular reaches this document only through a report citing an analysis that could not be opened. This is a Path A output, research-based definition, so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range 213,500,000–213,500,000 As of 2025 population estimates; framework status as of 2026-08-20

Sensitivity The interval has zero width and is not a confidence interval. It is the arithmetic sum of four published national population estimates, and no source opened here attaches a margin to any of them. The count measures residence, not harm: it counts everyone living in a country that entered the mechanism, not the people who bore the cost of a case staying open, and no source opened here distinguishes those two groups. It also mixes outcomes, because as of 2026-08-20 one of the four cases had produced a completed exchange, one had produced a treatment that rescheduled maturities without immediate debt reduction, one had completed only the official creditor track, and one was still at the stage of agreement in principle 66 months after application. Two corrections would move the figure in opposite directions and neither can be computed from the sources opened. Downward, because the count includes the two countries whose treatments were completed, so it overstates the population currently waiting. Upward, because 73 countries are eligible to apply and 78 IDA-eligible countries carried 1.2 trillion dollars of external debt as of December 2025, and the combined population of either group is absent from every source opened here. A far larger frame exists and is deliberately excluded: the United Nations reported in July 2023 that 3.3 billion people live in countries spending more on debt interest than on health or education, which is a measure of sovereign debt burden in general and not of this mechanism.

Regional breakdown The four countries are on three different sides of the same mechanism and the only sub-national data that would matter, which is who inside each country bears the cost of an unfinished restructuring, is absent from every source opened here. Splitting the national totals by region would be proportional allocation, and the burden of constrained public spending is not distributed with population. No geographic breakdown is asserted.