Sovereign restructuring was built for a world where delay cost the creditor nothing

The architecture that handles sovereign debt restructuring cannot handle the next wave of distress. That is my position. The reason is not creditor coordination, not the arrival of lenders who sit outside the old clubs, and not the drafting of collective action clauses. It is that the one form of relief this architecture reliably produces used to be free, and it has stopped being free.

What the process delivers, almost always, is time. Maturities are pushed out, grace periods are granted, coupons are stepped down and then stepped back up. Principal is rarely cut. That pattern is usually explained by creditor resistance to writedowns, and creditor resistance is real, but it is not the whole explanation. For most of the period in which the present machinery was assembled and rehearsed, an extension was a concession a creditor could sign without losing anything.

The concession that cost nothing

Take the discount rate everybody in the room is implicitly using. The ten-year Treasury constant maturity rate averaged 0.89 percent in 2020, 1.45 in 2021, 2.95 in 2022, 3.96 in 2023, 4.21 in 2024 and 4.29 in 2025. The series did not fall in a single year of that window. From 2020 to 2025 it rose 3.40 percentage points, which leaves the 2025 average at 4.8 times the 2020 one, and the largest single step was 1.50 points between 2021 and 2022.

Nominal yields are the wrong measure for a restructuring, because a debt stack is relieved in real terms or not at all. Set the ten-year breakeven against the ten-year yield and the picture gets sharper. Breakeven inflation ran 1.49 percent in 2020, 2.36 in 2021, 2.52 in 2022, 2.28 in 2023, 2.27 in 2024 and 2.33 in 2025. Subtracting one annual average from the other gives an implied real ten-year yield of minus 0.60 percent in 2020, minus 0.91 in 2021, 0.43 in 2022, 1.68 in 2023, 1.94 in 2024 and 1.96 in 2025. The swing across the window is 2.56 percentage points, and the sign changed between 2021 and 2022.

Now run the concession through both ends of that range. Defer a payment by ten years and discount it at the 2020 implied real rate of minus 0.60 percent, and its real present value rises by 6.2 percent. Defer the same payment by the same ten years at the 2025 rate of 1.96 percent, and its real present value falls by 17.6 percent. That is a gap of 23.8 percentage points on the identical concession. In the first case the creditor has given up nothing and has arguably gained. In the second the creditor has taken a writedown and has simply not called it one.

This is why the machinery looked serviceable. A voluntary process, whose enforcement runs no further than comparability of treatment and the reluctance of participants to be seen defecting, is well suited to arranging agreements that cost its members nothing. It was not tested on anything harder than that, and the years in which the concession was genuinely free are behind it. Extension produced headline relief for the debtor and no realised loss for the lender, and both sides could sign the same document for opposite reasons.

The exit assumption has not survived

Every reprofiling contains a bet that the borrower returns to the market before the deferred payments come due. That return is priced off the risk-free curve plus a spread, and the risk-free curve has not come back down. Six observations, no decline in any of them. What the series has done instead is stop climbing: the step into 2024 was 0.25 points and the step into 2025 was 0.08. A curve that has added 0.33 points across its last two annual steps and sits at 4.29 percent is not a curve that is reverting, and a deal that assumes reversion is a deal underwritten on a rate path nobody wrote down.

Currency compounds it where the debt is issued in dollars. The broad dollar index read 117.8 in 2020, 113.1 in 2021, 120.7 in 2022, 120.3 in 2023, 122.9 in 2024 and 122.7 in 2025. Against the 2021 reading the latest point is 9.6 index points higher, or 8.5 percent. That basket is not any single borrower's currency and I am not converting it into one. What it establishes is narrower. The index closed the window at 122.7 against 117.8 at the start, so the dollar is dearer against foreign currency in general than it was in 2020. It did not get there in a straight line: the index fell in 2021, again in 2023 and again in 2025. The direction across the window as a whole is still the one that makes a dollar obligation heavier to service out of local revenue.

The objection I take seriously

The best reply to all of this inverts it. If money costs more, holding out costs more too. A creditor funding a position at the rates now on offer has every reason to settle rather than litigate, and on that reading the rate reset strengthens the process instead of breaking it. The same reply adds that collective action clauses have in fact bound minorities where they were drafted into the paper, and that the newer official frameworks, slow and awkward as they are, have produced signed agreements.

Both halves of that are true and neither rescues the architecture. The funding-cost argument applies to a creditor who funds a position, which describes a distressed-debt desk and does not describe a bilateral official lender or an index fund holding to maturity. More to the point, higher rates change who wants to settle without changing who can be made to. They push creditors towards the settlement creditors prefer, which is deferral dressed as sacrifice, and that is precisely the settlement that no longer relieves anything once it is discounted properly.

The anchor is drifting too

There is a further complication, and it runs against the assumption that the discount rate will be obligingly stable while everyone negotiates. The country whose curve sets that rate is not holding still. Federal debt as a share of GDP fell for three consecutive observations, from 122.3 percent in 2020 to 120.9, then 118.4, then 117.7 in 2023, a total decline of 4.6 points. It then rose 2.7 points into 2024 and 0.3 into 2025, leaving the ratio at 120.7 and only 1.6 points below where the window opened. Restructuring terms are set against a benchmark issuer whose own balance sheet is moving, and no party to a Paris Club table controls it.

What I would change

One change would do more than another round of clause drafting. Every restructuring rests on a sustainability analysis with an interest rate path inside it, and that path is the single input every party can observe and nobody publishes. Publish it. State the assumed path, date it, and state in the same document what the agreed terms look like if the path is wrong by a hundred basis points in each direction. A deal that cannot survive that disclosure is a deal that was relying on the assumption not being read.

An admission is owed here, since this page argues about sovereign borrowers from four American series. The absence is part of the case. There is no comparable public record of the rate assumptions behind restructuring agreements, and a writer who wants to check one against the curve it was discounted at has nothing to check it with. The most transparent number in the whole arrangement is the one the arrangement never shows you.

So the question to put to the next agreement is narrow. Ask what real discount rate the relief was valued at, and ask when that rate was fixed. If the answer is a rate from the years when deferral was free, the agreement has moved payments and relieved nothing, and the wave it was meant to absorb is still coming.