For fifty years China absorbed the West's monetary expansion and sent back cheap goods to hide it. That capacity is now full. The adjustment has to land somewhere, and it is landing in the currencies themselves.
Volume I reconstructed a fifty-year monetary regime. Volume II asks what happens to that regime when the thing holding it up runs out.
Western credit and money expanded faster than the underlying domestic productive economy. China supplied the missing elasticity. Production moved outward, cheap manufactured goods moved back, and imported disinflation concealed much of the monetary expansion from the consumer price index.
China was not merely another trading partner. China was the absorber. Volume I put the mechanism this way: the productive capacity created in China allowed monetary expansion to continue while suppressing the politically explosive consumer-price consequences that would otherwise have accompanied it. The $200 trillion figure was the cumulative Chinese economic substrate that system ran through, not a $200 trillion bank transfer.
But an absorber cannot absorb forever.
By the mid-2020s China had become the world's manufacturing colossus while simultaneously confronting weak domestic demand, property contraction, debt, deflationary pressure and increasingly difficult employment conditions. So the question for Volume II is simple.
The hypothesis of this volume: the adjustment moves back into the currencies themselves.
The first transition was enormous because China began from a dramatically lower industrial and financial base. Western capital entered. Technology entered. Factories were constructed. Supply chains moved. Ports expanded. Hundreds of millions of workers entered industrial production. Western consumer markets absorbed the resulting output.
China's integration into the global economy created decades of additional productive elasticity. Volume I estimated that roughly 80% of China's cumulative 1970–2024 nominal GDP was generated after its 2001 WTO accession.
That transformation is non-repeatable.
The globalization machine consumed its own frontier. And the evidence of saturation is now visible inside China itself.
In its 2025 Article IV consultation, concluded in February 2026, the IMF describes weak Chinese domestic demand, deflationary pressure, a debt overhang, weakening productivity and "excess supply in some tradable sectors." Chinese authorities have their own word for it: involution, applied to iron and steel, glass, solar panels, electric vehicles, batteries and chemicals. The IMF also warns that exports may be less able to drive growth going forward.
That is the turning point. The absorber has become an exporter of its own surplus capacity.
The old Chinese problem was insufficient productive capacity. The new Chinese problem is insufficient demand for the productive capacity already created.
July 2026 shows the inversion in a single data release. Industrial production still grew 4.5% year over year. Retail sales grew 0.6%, missing a 1.5% forecast. Fixed-asset investment contracted 6.7% over the first seven months of the year, a steeper decline than the 5.7% drop recorded in the first half. Meanwhile the trade surplus came in at $112.5 billion for the month alone, part of a $687 billion surplus over seven months.
The factories are running. The domestic buyer is not showing up. The difference leaves the country.
The arithmetic of the old arrangement was Western demand in excess of Western production, with China filling the gap. The arithmetic now runs the other way: Chinese production in excess of Chinese domestic demand, with the gap having to go somewhere else.
The IMF describes exactly this imbalance. Weak domestic demand combined with industrial-policy-supported production has increased China's reliance on manufacturing exports and generated excess supply in tradable sectors, with what the Fund calls "adverse spillovers to trading partners."
The same industrial machine that once absorbed Western monetary expansion is now attempting to export its own excess production into the rest of the world. The absorber has changed polarity.
A claim circulates that 200 million Chinese men are unemployed. The instinct is to throw it out. That instinct is wrong, and it is wrong for exactly the reason Volume I was about. The $200 trillion was not a wire transfer either. Dismissing it on those grounds would have thrown away the mechanism. The number was real. It was measuring something other than what the headline said it measured.
Same here. So do the walk.
The number is published, and it is China's own. At the end of 2021 the National Bureau of Statistics counted 200 million people in what it calls flexible employment — linghuo jiuye. Self-employment, part-time work, platform and delivery work. That is not unemployment. It is employment without a permanent contract, without an employer, and largely without social insurance.
And it has not stayed at 200 million.
Now put that beside what China publishes as its headline.
Two numbers describe the same labour market. 5.2%, and 320 million. One of them travels.
And this is not a Chinese trick. It is the standard construction of labour statistics everywhere, including in the country doing the accusing.
Every statistical agency publishes a narrow number and a broad one, and the narrow one is the one that gets quoted on television. The United States is not hiding U-6. It publishes it monthly. It simply is not the number in the headline.
What changes across the Pacific is not the technique. It is the multiplier. America's entire U-3 to U-6 gap is about six million people. China's flexible-employment pool is roughly fifty times that, on a labour force only four and a half times larger.
So the corrected claim is not "200 million unemployed men." It is worse, and more useful. Somewhere between 280 and 320 million people are counted as employed while holding no contract, no employer obligation and no insured position — and the number has been rising through exactly the years the buildout matured, the property sector contracted, and fixed-asset investment turned negative.
Which raises the structural question this section exists to ask. How much labour does an industrial economy actually require once its industrial buildout is finished? Automation, robotics, surplus factory capacity, weak construction, demographic contraction and slower domestic consumption all cut the same way. The old globalization cycle needed hundreds of millions of additional workers. The mature system does not, and the gig economy is where the difference is being parked.
Volume I established that the monetary expansion did not show up in the price of bread, because Chinese production absorbed it. It did not establish where the expansion went instead. That omission matters, because money is not a gas. It does not fill a room evenly.
Richard Cantillon worked this out in the seventeen-thirties. New money enters an economy at a specific point and moves outward from there. Whoever receives it first spends it at the old prices. Whoever receives it last spends it at the new ones. The same nominal quantity is a gain at the front of the queue and a loss at the back, and the difference is not a side effect of the expansion. It is the expansion, seen from the position of the people it passes through.
Apply that to the arrangement Volume I described. Credit is created and enters through banks, primary dealers, government counterparties and the holders of existing assets. Cheap imports keep consumer prices quiet, so the expansion does not surface where a central bank would be forced to notice it. It surfaces instead in the price of the things the first receivers already own.
That is the mechanism behind a sentence Volume I asserted and did not explain: domestic asset prices rise. They rise because that is where the money is, and the money is there because of who touched it first.
And then it stops working. If each round of credit lodges disproportionately at the front of the queue, each round buys less of the thing the policy was for. You need a larger injection for the same real effect, which enlarges the distortion, which reduces the effect again. The question is whether that shows up in the data. It does.
TCMDO) and BEA via FRED (GDP, GDPC1, GDPDEF).In the nineteen-sixties the American economy converted $1.61 of new credit into a dollar of new real output. By the two-thousands it needed $8.45. Five times the credit for the same real result, which is what a saturated Cantillon channel looks like from the outside.
And then the line falls. That is where an argument built to prove a point would stop reading, so read it properly.
Between 2020 and 2024 nominal debt rose from $87.0 trillion to $107.2 trillion, up 23%. Over the same four years the GDP deflator rose 19%. The real debt stock therefore grew 3.5%. Credit did not become productive again. The unit moved, and the existing stock was worth less when it arrived. The ratio improves because the denominator of the debt was devalued, not because the numerator of the output was earned.
Which is this volume’s thesis, arriving early and from a different direction. The debt overhang was not repaid and it was not grown out of. It has been, and is being, inflated down. The only decade in sixty years where the credit-to-output ratio improved sharply is the decade in which the currency moved.
So the timing question this volume opened with has a second answer. Not only that the absorber filled, which closed the route by which expansion could be hidden. Also that the monetary channel had already been run to the point where it delivered five times less than it once did, and the resolution underway is the currency itself.
Both routes are shut. An administration that still needs to move something has to reach for instruments that are not monetary at all. That is the condition under which a tariff stops being a trade policy and becomes the only lever in reach.
One limit on the chart above, stated plainly. It measures a magnitude, not a distribution. That five times more credit was needed per unit of output is on the page. That the benefit accrued to holders of assets near the injection, and the cost to holders of wages further from it, is Cantillon’s claim rather than this chart’s finding. The evidence for it is not here but in Volume I §viii, The Ledger, where the ownership concentration is set out: the top tenth holding roughly 70% of equity wealth, the bottom half holding essentially none of it. The 2020–2024 column also covers four years, not ten.
The conventional explanation says tariffs are protectionism. One country taxes another country's products. The target retaliates. Consumers pay more. Businesses reorganize supply chains. That description is correct, and incomplete.
From the perspective of the monetary system, a tariff performs a second function: it deliberately reduces the ability of cheap foreign production to suppress domestic prices.
Which is close to the exact inverse of the previous forty-year arrangement.
This is why the tariff conflict should not be read only as trade policy. It is simultaneously an attack on the mechanism that previously helped hide monetary debasement.
Which raises an obvious objection: tariffs also raise revenue, and a government with a $40 trillion debt has an ordinary fiscal reason to want them. That objection is strong. It is also, as of 2026, testable.
On 20 February 2026, in Learning Resources, Inc. v. Trump, the Supreme Court held 6–3, Chief Justice Roberts writing, that the International Emergency Economic Powers Act does not authorize the President to impose tariffs. If Congress means to hand tariff power to the executive, the Court said, it has to do so explicitly.
The fiscal consequences were immediate and large. The IEEPA tariffs had raised more than $160 billion. Unwinding them opens up to $175 billion in refunds. The Penn Wharton Budget Model put the forgone revenue at $1.4 trillion over 2026–2035, with future tariff collections cut roughly in half. Customs and Border Protection stopped collecting on 24 February.
Here is the part that matters for this inquiry. If the tariff were fundamentally a revenue instrument, losing its revenue should have ended it.
It did not.
On 20 July 2026 the administration issued three proclamations reimposing 50% tariffs on Canadian goods under Section 338 of the Tariff Act of 1930, a provision no president had invoked in the ninety-six years it had been on the books. Section 338 permits duties where a foreign country disadvantages US commerce relative to other countries, and it caps the rate at 50%. The rate chosen was the ceiling. USMCA origination does not exempt covered goods, a sharp departure from every other Canada tariff regime.
That is close to a natural experiment. A policy stripped of its stated fiscal rationale, and reconstructed within five months on entirely different legal ground, is a policy whose operative purpose was not the stated one. What survived the Court was not the revenue. What survived was the wall.
And there is a second-order effect that is easy to miss. The refunds and the halved collections did not reduce the deficit, they widened it. When US debt crossed $40 trillion on 18 August 2026, forecasters noted it had arrived months ahead of schedule, in part because of revenue lost to the invalidated tariffs. The legal defeat of the tariff accelerated the debt that the currency now has to absorb.
At the political level this looks like a fight, and it is one.
The United States imposes tariffs.
Canada retaliates.
Industries complain.
Governments announce support packages.
The cameras record the argument.
The specifics, as of this writing: the 50% Section 338 duties took effect at 12:01 a.m. on 22 August 2026, after three days of talks collapsed, covering roughly $20 billion of Canadian exports including wine, furniture, dairy, cement, clothing and hockey equipment. Prime Minister Carney announced dollar-for-dollar retaliation the same day, effective 8 September, targeting US steel, electronics, dairy, appliances, agricultural equipment, and pulp and paper.
Game theory asks a different question than the press conference does. Not who is right. Rather:
If both sides make integrated North American production more expensive, capital has to reprice the entire production map. Automotive supply chains become less efficient. Steel becomes more expensive. Intermediate goods cross borders at higher cost. Investment decisions are postponed or moved. Workers lose bargaining stability. Governments borrow to compensate injured industries. Nominal costs rise.
Every one of those consequences lands on an already heavily indebted monetary system. Political conflict becomes an engine of monetary repricing, whatever either government intends.
This is where Volume II departs from Volume I. Under the old arrangement, China absorbed much of the adjustment through productive expansion. Under the emerging arrangement, the currency can absorb it instead.
Consider the full menu available to an overleveraged sovereign. It can default explicitly. It can drastically cut spending. It can tax enough to extinguish the debt. It can allow interest rates to rise until capital clears naturally. Or it can repress interest rates and reduce the real value of its liabilities through inflation and currency depreciation.
Politically, the fifth path has overwhelming advantages over the other four.
No legislature votes for a 20% haircut on every bond.
No president announces that household purchasing power will be reduced.
The unit of account simply buys less.
And the bond is repaid exactly as promised. One dollar remains one dollar. It is the thing the dollar purchases that changes.
This stopped being theoretical during one week in August 2026.
The long end of the Treasury curve had been in what traders openly called a buyers' strike since late June. On 17 August the 30-year yield reached 5.34%, its highest level since 2007, under pressure from roughly $2 trillion annual deficits, heavy long-dated issuance, corporate borrowing for the AI capital-expenditure boom, and inflation that has sat above the Federal Reserve's target for five years.
Two days later the Treasury responded, not by letting the yield clear, but by buying the bonds.
The Treasury at least doubled the maximum size of its long-dated buyback operations, from $2 billion to at least $4 billion, running September through November, targeting exactly the maturities the market had refused. Currency markets read the implication immediately.
Reuters reported the concern in plain terms: if Washington will not let borrowing costs rise, will the dollar end up absorbing the adjustment instead? The worry is not outright monetary financing. It is that officials facing rising debt-service costs may prevent yields from reaching market-clearing levels through buybacks, shorter-dated issuance and similar tools that reduce the duration private investors must hold.
The dollar fell. Gold rose more than 3% on the announcement. Traders revived a phrase for it: the debasement trade.
The choice begins to resemble a simple pair. Higher interest rates, or a cheaper currency. The debt does not disappear either way. Only its real value does.
There is nothing mystical about this mechanism. A heavily indebted state wants nominal economic growth above the effective interest rate on its debt. If nominal GDP rises rapidly while financing costs are restrained, the debt burden falls relative to the economy.
That can happen through real growth. But real productive growth is difficult. Currency debasement is easier.
Inflation increases nominal GDP.
Asset prices increase.
Tax receipts increase nominally.
Wages eventually increase.
Old fixed-rate debt remains denominated in yesterday's dollars.
The debtor wins. The creditor receives full repayment in a weaker unit. That is financial repression, and unlike a default it can proceed without anyone declaring that anything has defaulted.
This is the apparent contradiction at the centre of the whole episode. Why would a government trying to lower inflation impose tariffs that raise prices?
Because price stability may no longer be the dominant objective. The system may need nominal repricing.
A tariff can simultaneously:
In isolation, a tariff is a tax. Inside an overleveraged monetary system, a tariff can become one component of a much larger reflation machine. The Section 338 revival is what makes this reading hard to dismiss: the instrument was rebuilt after its revenue was struck down, which means the revenue was not what it was for.
This is where simplistic geopolitical framing becomes misleading, and where this inquiry parts company with the conspiracy version of its own thesis.
China does not have to defeat the United States. The United States does not have to surrender. Carney does not have to secretly work for Beijing. Trump does not have to consciously execute a Chinese strategy. Those propositions require evidence nobody presently has, and this reconstruction does not rest on them.
Game theory does not require them either. Suppose three players occupy the board.
Now A raises the cost of imports. B retaliates in kind. A and B together make their own integrated production system more expensive. C already possesses enormous alternative productive capacity and a surplus it needs to place.
C does not have to orchestrate anything. C gains relative position simply because A and B are dismantling the mechanism that previously advantaged them.
This is the difference between a conspiracy and a game. A conspiracy requires coordination. A game requires only incentives.
Nothing above should be read as a claim that China is winning. China has its own debt, its own property crisis, its own regional-government financing problems, its own demographic contraction, its own overinvestment, its own underconsumption, and its own currency to manage.
And China is already participating in the currency adjustment. The IMF found that China's low inflation relative to its trading partners has produced real exchange-rate depreciation, which strengthened exports and pushed the current-account surplus to an estimated 3.3% of GDP in 2025.
Read that mechanism carefully, because it is the quietest part of the whole story. China did not need a dramatic devaluation. If Chinese prices rise more slowly than American prices, China's real exchange rate falls without any exchange-rate event at all. No announcement. No crisis. No headline.
This suggests a different picture of the next monetary phase. Not one currency collapsing while everybody else stands still. Rather, competitive relative debasement.
Every major bloc faces liabilities that become easier to service in cheaper money. The United States wants debt dilution. China wants export competitiveness and domestic debt relief. Europe wants fiscal sustainability. Canada wants housing and sovereign-debt stabilization. Japan carries government debt that admits of no other resolution.
No participant wants its currency to depreciate uncontrollably. But no participant wants to hold the world's hardest currency either, because the hardest currency imports deflation, makes exports expensive, and increases the real burden of domestic debt.
So the equilibrium becomes strange, and stable:
That is monetary harmonization through competitive adjustment. It requires no summit, no treaty and no agreement. Only the same incentive arriving at every finance ministry at once.
Currencies are normally measured against one another, which is precisely what disguises synchronized depreciation. If the dollar loses purchasing power while the euro, yuan, pound, yen and Canadian dollar lose purchasing power at similar rates, the exchange rates between them can look comparatively stable. Nothing appears to have happened.
Measure all of them against something that cannot be produced by keystroke and the picture changes. Gold is useful here not because it has mystical properties but because it is an external denominator.
And here honesty is required, because the gold chart in 2026 is not the clean vertical line this thesis would prefer.
Gold set an all-time record of $5,597 an ounce on 29 January 2026. It then fell 28%, to roughly $4,020 by late July. That drawdown is not a footnote and it is not being minimized here. Anyone who spent the spring insisting that gold was screaming the imminent end of the dollar was wrong for six months, in a way that cost real money.
What happened next is the part that bears on this inquiry. Through August gold rose about 10%, its best month since January, closing at $4,577 on the 21st after a week of nearly 5% gains, driven explicitly by the Treasury's bond-market intervention, the $40 trillion debt print and a sliding dollar.
None of that proves the thesis. It is, precisely, the kind of indicator the thesis predicts, arriving on the days the thesis says it should. That is weaker than proof and stronger than nothing, and stating the difference is the whole discipline.
The general point survives the volatility. The system appears stable when currencies are compared with one another. The instability becomes visible when currencies are compared with scarce assets.
Volume I reconstructed the first. Volume II proposes the second. The hinge between them is the moment the absorber stops absorbing.
The public debate asks who won the tariff war. That may be the wrong question entirely. The more useful one:
Every government can tell a coherent national story about this, and each story is true on its own terms. Trump says America is defending American industry. Carney says Canada is defending Canadian sovereignty. China says it is defending free trade. Europe says it is defending European industry.
Meanwhile, underneath all four stories:
The debt remains.
The productive overcapacity remains.
Asset valuations remain elevated.
Demographic pressure remains.
Government spending remains structurally high.
And the external disinflation machine that made the previous monetary system politically sustainable is breaking down.
The national arguments can be entirely real while simultaneously functioning as the political surface of a deeper monetary adjustment. Nobody has to be lying for that to be true.
A thesis that explains every outcome explains nothing. So here is what each reading predicts, written down in advance, in a form that can be checked against the record later.
That is the experiment. No faith required. No secret meeting required. Watch the balance sheets.
Visual direction after the oceanographic films of Jacques-Yves Cousteau and the crew of the R.V. Calypso, 1943–1996. The saucer is the SP-350 “Denise”. No affiliation; the debt is the point.