Three documents were published in the same ten days of June: Beijing’s white paper on global governance, the G7’s reply, and a Council on Foreign Relations essay on why rules-based orders fail. Read together, they confirm the thing my bloc data has been saying for five posts:the China pole is built on institutions and arms, not money. And an order nobody will pay for doesn’t get a successor. It gets erosion.

While the world was watching the wars in the Middle East and Ukraine, the State Council Information Office published a 37-page document with the title More Just and Equitable Global Governance: China’s Principles, Proposals and Actions*. It is Beijing’s bid to be the architect, not just a participant, of the next international order. The white paper claims the project already has a constituency:

Upon its introduction, it swiftly gained support from nearly 160 countries and international organizations, with over 60 countries joining the Group of Friends of Global Governance. [link added]

The same week, the G7 published its own statement of how the world should run, and Benn Steil at the Council on Foreign Relations published an essay, Why Rules-Based Orders Fail, arguing that the whole idea of a rules-based order resting on its own rules is a category error. Put the three side by side and you get something better than any one of them: a primary-source confirmation of the central finding of this whole series, written by the party it should embarrass.

I have spent five posts arguing that the world has sorted into two blocs that are built very differently. The American bloc binds on security and finance. The Chinese bloc binds on trade and institutions and arms sales, and conspicuously not on money. In the bloc paper, when I ran the alignment data through a principal-components decomposition, arms imports loaded at 0.69 on China’s axis and US Treasury holdings loaded at −0.07. Translation: tell me a country buys Chinese weapons and joins Chinese institutions and I can place it. Tell me where it keeps its reserves and I learn nothing about its China tilt, because the answer is almost always “the dollar.” I called this the Eichengreen null, after his work showing that financial blocs in the 1930s followed the flag rather than leading it.

The white paper is Beijing writing down the Eichengreen null hypothesis in its own hand.

A 37-page bid to run the world that asks for a bigger seat at someone else’s table

A document about reshaping global governance, from the country that supposedly wants to displace the United States, does not propose a single major new financial institution, currency, or transfer. What it proposes instead is more votes inside the institutions that already exist:

the World Bank should conduct shareholding reviews and the International Monetary Fund should carry out quota share realignment in accordance with the agreed timeframes and roadmaps to address the democratic deficit in global financial governance.

Read that as what it is: China is not building a rival to the IMF, it is asking for a larger share of the IMF. On trade it goes further and volunteers to give something up:

… has announced that it will not seek new special and differential treatment in current and future negotiations at the WTO.

And it locates the problem with the present order not in the rules but in their enforcement:

Confrontation and injustice in today’s world do not arise because the UN Charter is outdated, but because it is not effectively implemented.

This is a reformer’s document, not a revolutionary’s. It wants a bigger seat at the dollar table, not a new table. Yu Jie at Chatham House, in the sharpest of the week’s commentaries, put her finger on what is missing:

It speaks extensively about principles, cooperation and institutional reform. But there are no major new financial commitments to help realize these ambitions.

She is right, and you can put a number on it.

Normative power, quantified: about four hundredths of one percent of GDP

Every architect of an international order has paid for the privilege. The Marshall Plan ran about $13.3 billion from 1948 to 1952, roughly 90% of it grants — close to 5% of a single year’s American GDP, better than 1% of GDP a year for four years. That is what underwriting a bloc looks like.

So I did the arithmetic on what China actually spends across borders on public goods, using the white paper’s own figures where it gives them and verified assessments where it doesn’t. The recurring, grant-equivalent total comes to something like $6 to $10 billion a year:

Against a Chinese economy of about $19 trillion, that is 0.03 to 0.05% of GDP. Even if you throw in the gross flow of New Development Bank lending, $42.9 billion cumulative since 2015, which is loans at roughly market rates in dollars, not transfers, you stay under a tenth of a percent. For comparison, even the United States, in the middle of gutting its own aid agency, still runs foreign assistance at something like 0.22% of GDP. The Marshall Plan was twenty-five to forty times China’s current intensity. Even diminished American aid is five to seven times it.

Yu Jie’s framing is the correct one, and it is worth quoting at length because it is the whole argument in three sentences:

Viewed through this lens, the white paper is less about financial hegemony and more about projecting normative power... Beijing is not proposing a Marshall Plan 2.0, nor is it offering to underwrite a global order through vast financial transfers or open-ended security guarantees... What remains uncertain is whether Beijing is willing — or able — to match its growing normative ambitions with the economic statecraft needed to realize them.

This is exactly the picture the financial-bifurcation posts in this series have been building. In Post 5, Dollar Debts, Yuan Revenues, the leverage China holds over the developing world runs through dollar loans it made through the Belt and Road: eighty-seven percent of that lending was denominated in dollars, not yuan. In Post 6, the asset side of the system would not move even after the largest reserve freeze in sixty years, because there is no liquid alternative that escapes the same jurisdiction. China spends 0.04% of GDP abroad and lends in dollars. That is not the balance sheet of a country building a financial bloc. It is the balance sheet of a country that has decided not to.

The G7 answered in the language of security, because that is the language it speaks

If the white paper tells you what the China pole is made of, the G7 statement tells you the same thing about the Western one by what it chooses to talk about. The document runs roughly 85% security and 15% economics. Ukraine, the Trump-brokered Iran deal, the Strait of Hormuz, Lebanon, Gaza, North Korea … and then one paragraph, near the end, on the world economy.

The binding clause is about China without naming it as the adversary:

We reaffirm our opposition to any unilateral attempts to change the status quo, in particular by force or coercion, in the East and South China Seas and across the Taiwan Strait, which should only be resolved peacefully through dialogue.

That is the Western bloc drawing its boundary on the security dimension, which is precisely the dimension my data says the Western bloc binds on. The one economic paragraph is almost plaintive by comparison:

We reaffirm our common interest in converging with other large economies on the causes of large and persistent global imbalances and on the need to address them.

“Large and persistent global imbalances” is diplomatic for China’s surplus, and the G7 punts it to the G20 under the American host year. This is the cooperative-rebalancing path I flagged in the trade posts as the alternative to bifurcation: the G7 would rather negotiate the imbalance down than sort away from it. Two poles, each binding on a different dimension: the West on security, China on institutions, and both of them only contesting economics rather than committing to break it. The domain divergence I built Post 2, How Countries Pick Sides, around is right there in the two documents’ tables of contents.

Why neither side can anchor the order it claims to want

This is where Steil’s essay does work the other two can’t. His piece, Why Rules-Based Orders Fail, is not about China at all. It is about a deeper problem, and it happens to explain why the white paper’s whole project is built on a foundation of sand.

Steil’s argument leans on Gödel’s Incompleteness Theorems and Arrow’s Impossibility Theorem, which sounds like a stretch until you see where he takes it. A rules-based order, like any system elaborate enough to govern its own operation, eventually meets questions its rules cannot answer: who decides the exception, who interprets the rule about interpreting the rules. So in practice the order never ran on its rules. It ran on a power willing to stand both inside and outside them:

In reality, the system depended heavily on the one country capable of operating simultaneously within and beyond the rules. The erosion of American predominance, and the rise of China as a near-peer competitor, has shattered that sensitive equilibrium.

Big words, simple point. The rules never enforced themselves. A hegemon enforced them, and paid for them, and broke them when it judged it had to, and the order held as long as everyone believed the exceptions were temporary. That is the thing the white paper does not grasp. Beijing thinks the problem with the order is that the rules are applied unfairly and the cure is more inclusive rules. Steil’s second move, borrowed from Arrow’s impossibility theorem, says the cure is the disease:

as a rules-based regime becomes more inclusive and diversity of preferences increases, coherent and broadly legitimate outcomes become harder to sustain.

His example is the one that should give the white paper’s drafters pause. The General Agreement on Tariffs and Trade worked, it cut industrial tariffs from about 35% in 1947 to 4% by the early 1990s, partly because it excluded the Soviet Union and so held together a club of broadly compatible economies. Then the United States insisted on a universal World Trade Organization and “assented to China’s admission before it had demonstrated adherence to core market principles,” and the WTO has been paralyzed ever since by the collision of incompatible economic models. More inclusive, less coherent. The white paper’s central demand — universalize the order, give everyone a vote, let the Global South in — is, in Steil’s frame, a recipe for an order that cannot decide anything.

The order doesn’t get a successor. It gets erosion.

Put the three together and the conclusion is not the one either side is selling. China is not the rising hegemon about to take the keys; it spends 0.04% of GDP abroad and asks for a bigger share of the dollar institutions it supposedly wants to replace. The G7 is not a confident bloc; it is assertive on security and almost apologetic on the economic imbalance that is actually driving the fragmentation. And Steil tells you why there is no clean handoff coming: the order was never the rules, it was the hegemon behind them, and the hegemon is now invoking the rules selectively itself.

The institutions remain, but they have been drained of authority. The rival major powers invoke rules selectively, interpret them opportunistically, or ignore them outright. The exceptions are no longer hidden in the background... They have moved to the foreground. And once seen, they can never be unseen.

This is the deep reason the contested middle of my alignment maps is metastable rather than sorting. I have made the point from the flows side, that the dollar architecture holds because, as the companion Three Clocks piece argues, the slow clock of reserve incumbency barely ticks while the fast clock of capital flight screams. The white paper is the same point from the other side. The architecture holds because there is no successor, and the white paper is Beijing telling you, in 37 pages, that it is not auditioning to be one. The renminbi sits near 2.3% of global reserves. China lends in dollars. It wants IMF votes, not an anti-IMF. A weak dollar with no replacement is not the changing of the guard. It is the guard staying put for lack of a relief.

But “no successor” is not the same as “no risk,” and this is the part that should worry you. The interwar parallel that opens the bloc paper, — the one where the open system of the 1920s fragmented into discriminatory blocs in the 1930s, did not feature a clean succession from sterling to the dollar either. It featured both incumbents and challengers invoking the rules when it suited them and abandoning them when it didn’t, until the rules meant nothing and the blocs hardened by default. That is the Steil scenario, and it is the one the three documents jointly describe: not China winning, but everyone defecting from the rules at once while the institutions stand hollow. An order erodes faster than it is replaced. You do not need a new hegemon to lose the old order. You only need the old one to stop underwriting it and the challenger to decline the bill — which is precisely what these three documents, read together, show both of them doing.

Three things, each a test of whether the erosion reading or the succession reading is right:

1. Whether China ever writes a check that contradicts the 0.04%. A genuine bid for financial leadership would show up as a real number — a development institution at Marshall-Plan scale, yuan lending that displaces the dollar loans, a reserve facility anyone actually draws on. So far, every Chinese financial move is denominated in the system it claims to be reforming.

2. The “global imbalances” paragraph. If the G20 under the American host year actually convenes a rebalancing negotiation with Chinese participation, the cooperative path is alive and the blocs are contesting economics, not breaking it. If it dies quietly, that is a tell the other way.

3. Which way the ~160 endorsers lean when it costs something. Endorsing a white paper is free. The Group of Friends of Global Governance can sign Beijing’s document and keep their reserves in Treasuries and their security under American guarantee, and most of them do. The number to watch is not how many endorse the vision but how many move a material variable — reserves, arms, a base — to match it.

TL;DR

China just published its plan to lead the world and forgot to fund it, because funding it was never the plan. The bid is for normative power on the cheap: more votes, more institutions, more “Groups of Friends,” and about four hundredths of a percent of GDP. That confirms what the data in this series has said all along: the China pole is institutions and arms, not money, and the dollar architecture holds because nobody is paying to replace it. The danger is not that Beijing wins. It is that Washington stops underwriting the order, Beijing declines to pick up the check, and the rules quietly stop meaning anything while the buildings still have the old names on them. That is not a handoff. That is the 1930s.

A short piece by Robin Harding, Why sinodollars outweigh the petroyuan (FT), does something useful: it cuts through the petroyuan noise with a single analytical distinction that I think is exactly right, connects directly to four years of prior work in this substack (warning: potential confirmation bias here), and illuminates something important about where the dollar’s structural position actually rests.

The concept is worth examining carefully, because it both validates part of what I argued in 2023 and complicates what I argued in late 2025.

The Petroyuan Lacks the Essential Property of the Petrodollar

Harding’s central move is a distinction most commentary on this topic blurs: settlement is not savings*

What made the petrodollar systemically important was not that oil was invoiced in dollars. It was that Saudi Arabia and the Gulf states held the dollars they earned, reinvesting them into dollar assets and creating the offshore dollar liquidity pool that finances global trade, sovereign debt, and aircraft leasing to this day. The petrodollar is best understood not as a payment, but as a dollar of oil profits in search of a home.

The petroyuan, by this standard, doesn’t exist in any meaningful sense. Today’s yuan earners don’t accumulate yuan savings, they spend them. Russia converts its yuan oil revenues into drone and ATV imports. Iran, facing reconstruction after the war, will spend every yuan it earns. Gulf producers no longer run the kind of structural surpluses that created investable pools in the first place. As Harding puts it:

This is a good backdrop to increase the use of the yuan in trade settlements, which is happening fast, but not to create a pool of offshore petroyuan.

Yuan settlement of China’s goods trade has reached 33.5 percent of total in March–April 2026, a record, per PBoC data. That is a real development. But 33.5 percent settlement share is a measure of invoicing and clearing. It says nothing about where the proceeds are parked.

When Saudi Arabia began recycling petrodollar earnings in 1975, the mechanics were constrained by market depth, not by political choice. David Mulford, who ran the Saudi Monetary Agency’s reserve portfolio in those years, later explained why alternatives to dollar assets were effectively unavailable at the required scale: purchases of German bonds, Japanese yen bonds, or Swiss franc notes “were just not possible in the sizes common in the U.S. market.” Settlement created savings in dollars not by design but because no other market could absorb the volumes involved. Today’s yuan earners face the same arithmetic in reverse: they settle in yuan and immediately spend it, because no market at sufficient depth exists to hold yuan savings at scale.

The more important monetary force, Harding argues, is what he calls the sinodollar: China’s relentless accumulation of dollar assets, driven by its current account surplus. China’s economic model requires running a surplus and recycling it into dollar assets. In this reading, the sinodollar is not a rival to the petrodollar, it is its structural successor as the dominant offshore dollar-creation mechanism. The world’s largest goods exporter is simultaneously the world’s largest dollar accumulator.

This Is the Pettis Argument, From the Other Side

For readers who have been following this series since 2023, Harding’s sinodollar concept should ring a bell. In the April 18, 2023 Perspective, I introduced a framework from Hyun Song Shin, articulated in his Odd Lots appearance, to evaluate the threat to dollar hegemony. Shin argues that for a currency to displace the incumbent, it must climb a ladder of reinforcing roles:

1. Invoicing transactions

2. Trade financing

3. Investing and borrowing

4. Currency hedging

Each step creates demand for the currency that makes the next step easier. Most challengers fail because they stall at Step 1: increased invoicing just leads to swapping the new currency back into dollars, because no one wants to hold the alternative. In 2023, I cited Michael Pettis making the same point with characteristic directness:

For the world meaningfully to switch from dollars to RMB, exporters will have to want to hold their accumulated surpluses in RMB and, much more importantly, China will have to give up control of its monetary policy and abandon its surpluses for permanent deficits. It is extremely unlikely that Brazilians will accumulate RMB assets in exchange for its surpluses.

Harding is making the same argument, from the opposite direction. Pettis said: foreigners won’t hold yuan because they prefer dollar assets. Harding says: China itself won’t stop accumulating dollar assets, because its growth model depends on the surplus. The two constraints are two sides of the same argument; the yuan can’t become a savings currency while both the issuer and the recipient prefer dollars for savings.

That argument still holds. But the picture has moved since 2023, and it has moved in a specific direction that Harding’s framing doesn’t fully capture.

The Ladder Has Been Moving — Just Not Where Harding Looks

By November 2025, the analysis in this newsletter had shifted. The November 26, 2025 Perspective documented a wave of data suggesting the RMB was advancing from Step 1 to Step 3, not through the conventional route (exporters choosing to save in yuan) but through a different mechanism: sovereign debt conversion.

Ethiopia was negotiating to convert at least part of its $5.38 billion in Chinese loans into yuan-denominated debt. Kenya had already completed a similar swap, saving $215 million a year in interest costs. Sovereign borrowers like Indonesia and Slovenia announced plans to issue renminbi bonds. Chinese trade credit, which ran 17 percent RMB-denominated as recently as 2021, had flipped to 72 percent RMB as of 2024.

The key insight from that analysis:

If Ethiopia owes debts in RMB, it must earn RMB to service them. This creates a structural demand for the currency that mere trade invoicing does not.

This is a fundamentally different pathway than what Harding’s framework addresses. The sinodollar story is about whether yuan earners will hold yuan savings. The liability-side story is about whether yuan debtors must earn yuan to repay. These two mechanisms are structurally independent. The sinodollar can remain intact — China accumulates dollars, China-bloc borrowers service yuan debt — and both can be simultaneously true.

That is, in fact, what appears to be happening. The sinodollar is real. The liability-side yuan bloc is also real. They are not in contradiction; they are operating on different parts of the monetary system’s balance sheet.

A parallel channel is visible in Chinese commercial bank lending. Dollar loans from Chinese institutions overseas fell from $587 billion in 2022 to $375 billion by late 2025, while yuan-denominated loans rose to $357 billion: the two now nearly equal, a crossing that was unthinkable five years ago. As Karthik Sankaran of the Quincy Institute wrote in FT Alphaville in February 20261, the real obstacle to renminbi bond market internationalisation “may be less the absence of the rule of law and more the absence of the rule of accounting.” That constraint narrows as Chinese lenders standardize disclosure and as borrowers convert dollar obligations into yuan. The BRI’s dollarization is unwinding from within, one refinancing at a time, through a mechanism Shin’s ladder framework did not anticipate.

The Shadow Fleet Complicates This Further

In the January 30, 2026 Perspective, I argued that the shadow oil trade had revealed a third pathway, one that bypasses Shin’s ladder entirely. The shadow fleet’s closed loop pairs yuan invoicing, yuan settlement through CIPS, yuan-denominated trade financing, and insurance through Sinosure, all within a single self-contained system. The argument was that this system “no longer needs to climb Shin’s ladder” because it operates as a closed loop outside dollar clearing.

Harding’s piece implicitly addresses this, and I think he is right to push back. A closed loop can function without climbing the ladder. But it cannot create the offshore savings pools that give a currency reserve status. Russia doesn’t hold yuan; it spends yuan. The closed loop is a transaction architecture, not a savings architecture. The shadow fleet proves the renminbi can work as a payment medium in bilateral sanctioned trade. It does not prove the renminbi can become a store of value at systemic scale.

Both observations are correct at different levels of analysis. The closed loop is real and growing:CIPS hit a record 921 billion yuan daily in March and briefly exceeded 1.22 trillion yuan per day in April following the Iran war shock, two consecutive monthly records. As Bert Hofman, former World Bank country director for China, observed of the shadow fleet’s structural yuan dependency: “Russia can’t use dollars. There’s nothing much else out there that would not have to hit the dollar system at some point.”2 The 193 institutions now bypassing SWIFT entirely, up 40 percent since 2024, are not choosing yuan as a preference. They are held in it by the same sanctions architecture that removed the dollar option. But Harding is right that this activity, at its current scale, does not build the reserve pool that the petrodollar built. It is a functional bypass; it is not yet a structural replacement.

The Sinodollar Is the Last Structural Prop

In the April 30, 2026 Perspective I argued that the dollar’s incumbency in 2026 is being held up by supports that are structurally different, and more contingent, than what held it in place a decade ago. The foreign official sector held 46 percent of Treasuries in 2008; it holds less than 30 percent today. Hedge funds in the Cayman Islands absorbed 37 percent of net Treasury issuance in recent years. Dollar SSA spreads have compressed to near-zero as a Treasury substitute. The Swiss franc hit a 14-year high against the dollar driven by foreign asset managers adjusting hedge ratios: not de-dollarization, but a mechanical unwind of passive dollar longs implicit in unhedged S&P holdings.

Each of those is a real market mechanism. None is structural in the way that foreign official demand for Treasuries was structural for the previous generation.

What Harding’s piece adds to that analysis is the identification of the one remaining structural prop: the sinodollar. China’s CA surplus recycled into dollar assets is the support that the contingent props are sitting on top of. It doesn’t show up in COFER because Brad Setser has documented that China moved its dollars — SAFE’s formal reserves fell, but state commercial banks absorbed over a trillion in dollar claims, the policy banks hold another trillion in dollar lending, and the China Investment Corporation holds $450 billion. As Setser put it: “Nice little trick. It seems to have fooled most of the internet.” The dollars are there. They just changed nameplates.

A May 2026 FT piece, America needs to put the renminbi back on the international agenda, by Mark Sobel, Brad Setser, and Robin Brooks put the system’s scale in direct terms: China’s manufacturing surplus is “close to 1 per cent of world GDP — a much bigger surplus than any single country has run in the last 70-plus years.” On their assessment, the reported current account surplus of 3.7 percent of GDP in 2025 understates the true figure, with state commercial bank purchases of dollar assets serving as the de facto sterilization mechanism. Setser’s February 2026 FT piece documented the operationalization: state banks purchased an estimated $100 billion in foreign exchange in December 2025 alone, a record monthly figure, with another $70 billion in January 2026. These are not passive accumulations. They are deliberate interventions designed to prevent RMB appreciation from compressing the surplus model that generates the sinodollar in the first place.

The sinodollar is why the BIS wave thesis holds. It is why, in the April analysis, “the movement remains dollar” even as the bezel gets more crowded.

One Thing Harding Does Not Address

Harding frames the sinodollar as a durable feature of China’s model. I think this is the one place where the analysis needs to be more careful.

China has been a consistent and large buyer of gold for its central bank reserves. This matters because the chain Harding describes — export surplus, earn dollars, recycle into dollar assets — is being partially interrupted at the savings step. A portion of China’s dollar earnings is not being recycled into Treasuries or dollar-denominated claims. It is being converted into gold: a neutral reserve asset, outside any currency system, immune to the kind of freeze that immobilized Russia’s dollar reserves in 2022. On official PBOC disclosures, the conversion rate runs in the low single digits of the annual surplus, with 2023 as a peak year where even the official figure touched the mid-single digits, and evidence that purchases routed through state commercial banks and the Shanghai Gold Exchange run the true rate higher. The direction is deliberate and the pace has accelerated since 2022.

This is not, by itself, a reversal of the sinodollar mechanism. China is still the world’s dominant dollar recycler. But the sinodollar prop has a leak in it, and the leak is the PBOC’s gold desk. The accumulation is not speculative. China holds no US dollar swap line; no Federal Reserve backstop for dollar funding stress. In a reserve framework where buffer value depends critically on network access, gold is the rational self-insurance instrument for a country that is simultaneously the dollar system’s largest structural prop and excluded from the dollar system’s liquidity architecture.

More significantly, the sinodollar has a demographic expiration date. China’s working-age population is contracting. The current account surplus that generates the sinodollar flow is likely near its structural peak. The same demographic forces that our bloc formation work identifies as reshaping geopolitical alignment over the 2030–2044 window will, by compressing China’s surplus, also compress sinodollar creation. The structural prop that is currently holding up the dollar’s incumbency is time-limited in a way that Harding’s framing does not acknowledge.

Summing Up

Harding’s sinodollar concept is the clearest short-form articulation of why dollar dominance is more durable than the petroyuan narrative implies. Settlement is not savings. China’s own model has been, and for now remains, the largest source of structural demand for dollar assets. The 33.5 percent yuan settlement share is real; the reserve accumulation it implies is not.

But the concept also does exactly what good analytical framing does: it clarifies where the real vulnerabilities lie. The dollar’s structural prop is not the Fed’s liquidity backstop, or the Treasury market’s depth, or the SWIFT network’s inertia. It is China’s CA surplus, and China is using that surplus in ways that Kissinger’s Saudi recycling model did not anticipate. Some of it goes into gold. All of it rests on a surplus that will compress.

The analysis in this substack has been that the ladder is being climbed, step by step, through mechanisms Shin’s original framework did not fully anticipate: closed-loop shadow settlement, liability-side debt conversion, and now the gold drain on the sinodollar mechanism itself. Harding’s piece doesn’t change that conclusion. It sharpens the picture of what is holding the current equilibrium in place, and how contingent that equilibrium has become.

Benguria, Rojas and Saffie (BRS) have recently published Geopolitical Fragmentation, Sovereign Debt, and Dollar Dominance.

• Side note: I’ve posted Dollar Debts, Yuan Revenues: Testing the BRS Paper Against Our Demographic Framework over on the Demographics & Capital Flows substack that analyzes the paper through the demographic, rather than geopolitical, lense.

The first four posts in this series posited that the world has sorted into two blocs over twenty years, that Chinese development finance is the most powerful time-varying predictor of which way countries go, that the countries most economically dependent on China still route their financial surpluses through the dollar system, and that this trade-capital divergence is a structural feature, not a transitional lag that demographic gravity alone will not resolve.

What none of those posts addressed is the question underneath the question: if the sorting continues and the mismatch between where goods trade and where money flows keeps widening, what is the specific mechanism by which the financial side eventually gives way? Post 3 ended on the word “metastable” and noted that the transition, when it comes, is triggered by crisis, not drift.

The BRS paper is the first formal model I have seen that identifies a concrete liability-side mechanism for that crisis: not a sudden flight from Treasury securities by central banks, not a collapse in dollar FX volume, but something quieter and more structural; sovereign borrowers converting the debt that China’s own lending program created into the currency that China’s own trade relationships are making more attractive. The BRS paper is worth taking seriously because it closes the loop that Post 3 left open.

Dollar Debts, Yuan Revenues

In March 2025, Kenya converted $4.9 billion in Chinese railway loans from dollars into yuan, saving roughly $215 million per year in interest costs — 4.3 cents on the dollar of face value. The conversion was quiet enough that it barely registered in the financial press. It should not have been quiet.

The Kenyan railway deal is the first concrete data point in what a team of economists at the University of Kentucky, University of Florida, and University of Virginia have formalized as a sovereign debt restructuring cascade. BRS describe a mechanism by which geopolitical fragmentation could erode dollar dominance not from the asset side of sovereign balance sheets, what central banks hold in reserves, but from the liability side: what governments owe, and in what currency.

The asset side receives most of the empirical attention, which is why the BRS paper is worth taking seriously. The liability side is where the map is actually moving.

I spent the last several weeks running their framework against our 46-country geopolitical alignment data from this series. This post is what I found.

China Lent in Dollars

The counterintuitive fact at the center of the BRS argument: the loans China extended through its Belt and Road Initiative and state policy banks are overwhelmingly denominated in dollars. Not yuan. Dollars. Eighty-seven percent by commitment value, according to the AidData dataset tracking Chinese overseas lending from 2000 to 2021.

Think about what this means structurally. China’s principal instrument for building economic dependence across Africa, Asia, and Latin America created dollar liabilities for the recipient countries, not yuan liabilities. Kenya did not borrow yuan to build a railway. Ethiopia did not borrow yuan to build industrial parks. Angola did not borrow yuan to develop oil infrastructure. The net effect of a decade and a half of Chinese development finance was to deepen the developing world’s exposure to the dollar system, not to replace it.

This is the setup for the mismatch. Fragmentation has redirected trade toward China and yuan-linked markets. Kenya’s exports to China grew from 9% of its total trade in 2010 to 16% in 2024. Ethiopia’s from 19% to 31%. Angola’s China trade share sits at 34%. As trade reorients toward China, the revenues that service those dollar loans become increasingly yuan-denominated at origin, even if they clear in dollars. The friction rises. The mismatch is the structural result.

The Mismatch Zone

The BRS framework identifies the mismatch zone as countries where the dollar share of sovereign debt exceeds their effective dollar export-revenue exposure. These are the countries below the 45-degree line in their diagram: dollar debt greater than dollar invoicing. They face a rising real cost of debt service as fragmentation continues, and they are the candidates for following Kenya.

I do not have the Boz, Casas, Diez, Gopinath, and Gourinchas (Patterns of Invoicing Currency in Global Trade, IMF WP 2025/178) data locally, and that gap matters; I will come back to it. Using the China trade share as a proxy for yuan revenue exposure, and merging World Bank International Debt Statistics on the dollar share of public external debt for the 24 developing-economy borrowers in our 46-country sample, seven countries fall clearly into the BRS mismatch zone with meaningful yuan exposure alongside high dollar liabilities:

These are the countries structurally positioned to do what Kenya did: convert existing dollar debt to yuan if yuan refinancing becomes cheaper.

Notice what is not on that list. Kenya itself does not fall below the 45-degree line in the data. Its dollar debt share is 73% and its China trade proxy is 16%, dollar invoicing exceeds dollar debt, by this crude measure. That is not a failure of the framework. It is a finding. Kenya converted not because it was the most structurally exposed country in the sample but because it found a specific deal at a specific moment with a specific creditor. The structurally exposed countries are in the table above, not in Kenya.

The Problem With Angola

Angola belongs at the top of that table in multiple ways. It has the highest cumulative Chinese development finance in our sample, 0.63 times its average annual GDP, accumulated over 2000–2021. Its dollar debt share is 87%. Its China trade proxy is 34%.

But Angola is also the clearest illustration of why invoicing data matters more than trade shares.

Angola exports crude oil. Oil accounts for roughly 60% of its exports. Crude oil is priced in dollars globally and invoiced in dollars regardless of whether the tankers go to Rotterdam or Qingdao. A Luanda exporter selling crude to Sinopec receives dollars, not yuan. The effective yuan revenue exposure is not Angola’s 34% China trade share. It is closer to 34% times 40%, adjusting for the oil tranche — roughly 14%.

At 14% yuan exposure and 87% dollar debt, Angola is still currency-mismatched. But the mismatch is meaningfully smaller than the headline number suggests, and the incentive to restructure is weaker than the table implies. The countries with the strongest restructuring logic are the non-commodity cases: Pakistan (textiles, remittances), Kazakhstan (more diversified than Angola despite mining dependence), the Philippines (electronics, remittances). Their yuan invoicing tracks closer to their trade share.

This matters for the cascade mechanism because that mechanism is hardest to trigger when countries are heterogeneous in their effective yuan exposure. Angola and Nigeria, the two largest African recipients of Chinese development finance, are both commodity exporters whose revenues remain more dollar-linked than their trade shares imply. The cascade is structurally weakest precisely where the lending went deepest. This is the main reason I would not read too much into the headline mismatch numbers without the Boz invoicing data.

A Cascade Condition That Is Not Met

Here is the formal structure of the Benguria et al. model. Countries vary in their yuan revenue exposure. Each country restructures its dollar debt into yuan when that exposure crosses a threshold, and the threshold falls as more countries restructure, because each restructuring deepens the yuan sovereign debt market, lowering refinancing costs for the next country. The math produces a tipping condition. Let h be the half-range of the cross-country distribution in yuan revenue exposure, a measure of how similar or different countries are. Let Λ be the liquidity feedback parameter, how much each restructuring deepens the yuan market. The cascade triggers when:

Λ > 2ℎ

When countries are similar enough in their yuan exposure, early restructurings can pull the rest across. When they are heterogeneous, the cascade fails.

From our 46-country data, h is large. The China-aligned cluster shows a half-range of roughly 0.13 in the trade-share proxy for yuan exposure, running from Turkey (8%) to Russia (35%). The full non-China sample extends further, to a half-range of 0.15. Under any reasonable h specification, the cascade threshold is 2h ≈ 0.18 to 0.30.

Kenya’s restructuring implies Λ at most 0.043. The 4.3% interest saving on that specific deal is an illustrative ceiling on Λ, not a market-depth estimate — Λ measures how much each restructuring deepens the yuan market for the next borrower, which is a different and structurally harder thing to measure from a single deal. Taken at face value: at Λ ≈ 0.043 and 2h ≈ 0.18 to 0.30, the cascade condition falls short by a factor of four to seven. Closing the gap from near-zero to the cascade threshold requires roughly four to seven additional Kenya-scale restructurings happening close enough together to materially deepen the yuan sovereign debt market. That is approximately the scale of a coordinated conversion wave across several HIPC-eligible African and Asian Chinese borrowers simultaneously.

Not impossible. Not imminent. Somewhere in the 5–10 year range on current trajectories.

Central Banks Are Not Adjusting on the Other Side

The liability side of the sovereign balance sheet is, slowly and case by case, starting to move. The asset side is not.

In the Chenard, Eichengreen, Monnet, and Morvillier paper (CEPR DP21488, May 2026) and in my own follow-up work on reserve composition, the result is consistent: the dollar’s reserve role is orthogonal to geopolitical alignment in central bank reserve management behavior. Countries that have drifted toward China — in their political alignment, their institutional memberships, their trade patterns — do not show lower dollar securities shares in their official reserves. Countries that have drifted away from China do not show higher ones.

I tested this directly using our composite alignment score as the predictor variable in a panel regression on securities shares, following the Chenard et al. specification. The 7-dimension alignment index (combining trade, capital flows, diplomacy, arms, development finance, and institutional membership) has no predictive power for reserve composition within countries over 2010–2024 after controlling for country fixed effects (coefficient +51 percentage points per unit, p = 0.27). It has no cross-sectional predictive power either (p = 0.46). The richer index fails where Chenard’s simpler UNGA-based proxy fails, and that is the result. Central banks compartmentalize. Their reserve management is driven by financial-system considerations — adequacy, liquidity, market depth — and not by the political and economic alignment decisions being made simultaneously across the rest of the government.

This is consistent with the larger H4 null from the reserve composition work: aging creditor countries did not rotate toward deposits after the 2022 Russian reserve freeze (β = −0.203, p = 0.570). The reason, I argued there, is institutional logic, deposits at the Bank for International Settlements face the same legal jurisdiction as US Treasuries. Within the dollar system, instrument switching does not reduce seizure risk. Genuine de-risking requires gold or non-dollar currency reallocation, not instrument switching within the system. Central banks know this. They are not, on average, doing it.

The Map Is Making the Mismatch Worse

The mismatch would be concerning if it were static. It is not static.

For each of the 24 developing-country borrowers in our sample, I computed the trend in China trade share over 2010–2024. Every single one of the seven BRS-mismatch countries shows a positive trend. None is drifting toward a smaller mismatch. All are drifting deeper into one.

Indonesia is the most striking. Its China trade share has risen at 1.2 percentage points per year. At that rate it crosses 39% by 2034, against a dollar debt share of 90%. Angola projects to 41% China trade by 2034. Brazil ran from 15% (2010) to 26% (2024) and projects to 36%. These are large middle-income economies, not isolated HIPC cases.

A second wave is approaching the threshold. Thailand, Bangladesh, Kenya, and Colombia are currently below the 20% China trade share that anchors the BRS-faithful mismatch condition. On current trends, all four cross it by 2034 while carrying dollar debt shares above 70%. Thailand at 93% dollar debt and 24% projected China trade share is an illustration of how quickly the second wave could materialize.

What This Means for Post 3’s Conclusion

In The Trade-Capital Divergence, I characterized the current international monetary arrangement as a metastable equilibrium: stable against small perturbations, vulnerable to a sufficiently large shock. The dollar system’s network effects and institutional depth make it self-reinforcing. Countries trade with China and finance through the US not because they chose this but because the plumbing routes surpluses into dollar assets automatically.

The Benguria et al. mechanism is the liability-side version of the same logic. Developing countries did not choose to borrow in dollars. They defaulted into it — because China, their principal bilateral creditor, offered dollars. The dollar system is self-reinforcing on both sides of the balance sheet simultaneously. Dollar assets accumulate because the plumbing intermediates everything. Dollar liabilities accumulated because the largest alternative creditor priced in dollars.

The tension is that both sides are now beginning to move, but at different speeds. The liability side moves through sovereign debt restructuring: case by case, one railway at a time, at the pace of distressed debt negotiations. The asset side is not moving at all in the central bank data. Individual country defections face prohibitive switching costs on both sides, but the switching costs are structured differently. Restructuring a specific loan is a bilateral negotiation with one creditor. Shifting reserve composition away from Treasuries requires replacing a market with no peer; $27 trillion in outstanding securities, 88% of global FX transaction clearing, 54% of global trade invoicing. The threshold for unilateral reserve reallocation is much higher than the threshold for restructuring a single loan.

This asymmetric stickiness is the specific risk. In a world where liabilities and assets adjusted together, where each yuan restructuring was accompanied by a corresponding shift in yuan reserve accumulation, the balance sheet would rotate gradually. But central banks are not doing that. The liability side shifts. The asset side does not. Each dollar loan that gets restructured into yuan deepens the structural logic for yuan reserve accumulation without triggering it. The balance-sheet divergence compounds quietly.

When that divergence eventually closes, it closes through the asset side catching up to the liability side. That process, when it comes, will not be gradual.

The Bottom Line

The cascade is distant. Four to seven more Kenya-scale restructurings, and then only if the yuan sovereign debt market deepens proportionally, which requires PBOC commitment to market infrastructure that does not yet exist at scale. The mismatch is real for seven countries but smaller than the headline numbers suggest, for the same reason that Angola’s trade with China does not make Angola a yuan-revenue economy: oil is oil.

What is not distant is the accumulation. The BRS-mismatch countries are getting deeper into the zone every year. Indonesia is adding more than a percentage point of China trade share annually. The reserve portfolio is not adjusting. The structural gap between dollar liabilities and potential yuan reserves is widening each year the sorting continues. Kenya converted one deal. Indonesia is building toward a much larger one.

The question I ended Post 3 with, whether the financial plumbing adapts gradually or all at once, now has a partial answer on the liability side: gradually, so far. The trend is not toward gradually.