AIERA FrontiersAIERAFrontiers
All articles

The Three Crutches of the Dollar: Who Really Holds Up the Global Financial System. Part 2 of 3

The dollar rests on three temporary structures: Treasury T-bills, the Fed's reserve 'floor', and the Bank of Japan's cheap yen. We break down each one — and how they could collapse.

AIERA FrontiersAugust 17, 202619 min

Key takeaways

  • The dollar rests not on gold or 'fundamentals' but on three temporary crutches: endless Treasury T-bills, the Fed's bank-reserve 'floor', and the Bank of Japan's cheap yen
  • Crutch #1: Yellen's 2023 maneuver — issuing T-bills instead of long bonds drained the ON RRP cushion from $2.25T to zero and 'outbid' Fed policy by ~0.25pp of yield
  • Crutch #2: the Fed hit LCLOR (~$2.7T, 8% of GDP) and halted QT — beyond it is panic, as in September 2019
  • Crutch #3: the yen carry trade ($1+ trillion) subsidizes demand for Treasuries; every BoJ hike hits it (5 Aug 2024 — Nikkei −12.4%)
  • Regulators are building Plan B — digitalization (RLN, wCBDC, tokenized Treasuries): margin calls in milliseconds, but more control and less privacy
Article podcast
aierafrontiers.com/en/article/tri-kostylya-dollara-en

The modern dollar architecture rests on three "crutches." Not figuratively — literally: three temporary structures, each born as an emergency measure. Together they work. Separately, they do not.

Remove any one — and regulators will have to do what they have long feared: print money directly to cover government debt. In textbooks this is called "monetization of fiscal deficit." In real life — inflation, devaluation, loss of trust.

Three points of support

These three crutches hold the dollar together. See what happens if you pull any one out.

CrutchWhat it holds up
#1 · T-billsThe US Treasury prints short bills to feed the shadow bank with collateral.
#2 · Fed reservesThe Fed keeps bank reserves above $2.7 trillion, refusing to shrink its balance sheet.
#3 · Yen subsidyJapan keeps rates low → global funds buy dollar debt.
Main thesisWe are at a point where any of the three conditions may break first. The Bank of Japan is tightening. Hedge funds sit on 100x leverage. The Fed is formally "independent," but its balance sheet is effectively tied to Treasury issuance. There are several break scenarios — and far fewer "soft landing" ones.

The key insight — this is not a ring: the Treasury sits in the center, held up by three props. And the props hold up each other (the red dashed arrows in the diagram). So if one falls, all fall.

  • MMF ↔ reserves: as long as money-market funds pull liquidity from the ON RRP into T-bills, bank reserves barely suffer.
  • Reserves ↔ carry: the yen carry trade supports demand for Treasuries, and with it the stability of reserves.
  • Yen → Treasuries: cheap yen loans feed demand for American debt.

Mental model: rehypothecation

The Appendix has the apartment analogy. Here are the numbers — without them you cannot see why a "shortage of paper" equals a thrombosis of the whole system.

StepWhat happens
1Hedge fund A owns a $100M Treasury and pledges it to prime broker B in a $100M repo.
2B reuses the same paper (rehypothecation): borrows $100M from fund C against that same paper.
3Fund C gives the paper to bank D and borrows another $100M.
Total1 paper = $300M of credit promises. Velocity = 3.1 (as in 2007).

This is the velocity from Chart II. Manmohan Singh measured exactly this: in 2007 velocity was 3.1; by 2020 it fell to 1.9 — the system became afraid to reuse.

Why it works — and why it is dangerousIt works because everyone believes the paper will return tomorrow (overnight repo). It is dangerous because there is one paper and three promises. In calm, the chain flips over every day. In panic, owner A demands the paper back: B gave it to C, C to D — and to return it B must find $100M of cash, C from D. The chain collapses instantly.
⚠️ Breaking pointIf Treasuries run short (and they do — everyone sits in them as collateral), at the first margin call nobody can find paper. This is not a "shortage of money" but a shortage of one specific paper promised to three people. When velocity falls from 3.1 to 1.9, the system loses $2.4 trillion of "artificial" money: total reusable collateral fell from $4.5T (end-2007) to $2.1T (end-2009, per Singh & Aitken).

Crutch #1. The endless T-bill machine

In 2023 the US government was stuck: Republicans in Congress refused to raise the debt ceiling, and default was weeks away. They struck a last-minute deal — on the condition that the Treasury "get creative" to avoid pressuring long rates.

Treasury Secretary Janet Yellen applied a trick described by economists Steven Miran and Nouriel Roubini back in 2022: instead of issuing long bonds (10–30 years), issue masses of short bills (T-bills, up to a year). The bills offered money-market funds a yield slightly above the Fed's reverse-repo rate (ON RRP) — and money from that liquidity "parking lot" flooded into T-bills. The Treasury got funded, and bank reserves barely moved.

One figureThe ON RRP "cushion" shrank from $2.25 trillion to almost zero in a year and a half (mid-2023 to end-2024). The Treasury drained the Fed's liquidity piggy bank.
The physics: how the Treasury outsmarted the FedBefore the move, funds hold $2.25T in ON RRP at 5.2%, and bank reserves are $3.2T. Yellen's move: T-bills yield 5.3% — 0.1% above RRP. Funds switch (nobody forces them — it is simply better). After the move, ON RRP = $0, the same $2.25T now sits in T-bills, and bank reserves are $3.25T (a ~5% change). Result: the Treasury is funded, the Fed formally lost nothing — but the cushion is gone, and the next T-bill will pull straight from bank reserves.

By Miran and Roubini's estimates, the maturity manipulation cut 10-year Treasury yields by about 0.25 percentage points — the equivalent of a 1% Fed rate cut. In other words, the Treasury "outbid" the regulator's own monetary policy. That used to be unthinkable.

Why it works (for now)

Short bills are ideal collateral for funds: short (low risk), dollar-denominated (zero currency risk), government-backed (zero credit risk). While T-bills sit with funds, they do not "press" on the long end of the curve — mortgage rates, auto loans, and corporate bonds stay lower than they would otherwise be. When you see "Treasury places $X billion of T-bills" in the news — that machine is running.

Analogy: you feed a cat allergic to cheap food. Every day you buy it expensive food with your own money, and the cat gets used to it. Money runs out — and you "feed the cat" with bills you issue yourself. It works as long as the cat keeps eating.

Crutch #2. The floor of bank reserves. Why the Fed no longer tightens

In 2008 the Fed's balance sheet was $900 billion, by 2014 — $4.5 trillion, by 2022 — $8.9 trillion. A printing press on full. In 2022 the Fed began "undoing" QE with its QT program: each month it stopped reinvesting interest, and money drained out of the system.

But each new T-bill issue (crutch #1) began pulling reserves straight out of banks. The ON RRP cushion that used to absorb that blow ran out at the end of 2024. After that — only falling bank reserves. And there the Fed hit its "lowest comfortable level of reserves" (LCLOR).

One figureBy 2025 the Fed had radically slowed QT: from $60bn/month in 2022 to $25bn/month in May 2024 and $5bn/month in April 2025. Officially — "reserves reached a comfortable level." Unofficially — beyond that point panic began.

What LCLOR means in plain words

Banks hold reserves at the Fed — the money they use to settle with each other daily. If there is too little, what happened in September 2019 occurs: overnight rates spike from 2% to 10%, the market "freezes," and the Fed injects hundreds of billions in an emergency. LCLOR is the level below which the system starts to "suffocate": roughly 8% of US GDP, or about $2.7 trillion.

Analogy: you have 100 rubles in the bank and a bill of 80 to pay. If the bill is 90 — tight, but you manage. If 110 — you don't, and the bank freezes the next operation. The Fed's "account" is the LCLOR limit. They have reached it, and the next dollar of deficit is a "bill of 110."

What this means in practiceThe Fed is formally "independent," but its balance sheet is effectively tied to Treasury issuance: as long as the Treasury prints T-bills, the Fed cannot shrink its balance sheet. This is a structural constraint, not politics — no new Fed chair can fix it. For you: more money stays in the system than in a normal scenario, so dollar savings lose purchasing power a bit faster — a few percent a year, which compounds visibly over a decade.

Crutch #3. The yen subsidy. While Japan keeps rates low, the world buys dollars

Remember the story from Part 1: since 1983 Japan has been pulled into the American financial system. Plaza 1985, the 1990 bubble, ZIRP, NIRP, YCC, a Bank of Japan balance sheet bigger than GDP. By 2024 Japan holds more than $1.1 trillion of US government debt — its largest foreign buyer.

Why? The yen is the "cheapest currency to borrow." You borrow yen at 0% (in 2016–2024 even at −0.1%), convert to dollars, buy Treasuries at 4–5% — the difference is pure profit. That is the carry trade. In 2023 alone its volume with the yen hit a record $1+ trillion. For the Treasury this is free demand for its debt: 10-year yields stay lower than they would be, and mortgages and autos are cheaper.

The physics: how carry works and where it breaksBorrow ¥15bn at 0.1% from the BoJ, convert at 150 ¥/$ into $100M, buy a T-bill at 4.3%. The spread is 4.2% = $4.2M a year "out of thin air"; with 10x leverage that is $42M of income on $10M of your own (420% a year on paper). Reversal: yen 150→135 per dollar (10% in a day, like 5 August 2024) turns a $100M debt into $111M — an $11M loss against $4.2M of annual income, three years of profit wiped out in one day. Margin call → selling Treasuries → yields rise → Nikkei falls → the yen strengthens further — a loop.
⚠️ Breaking pointEvery BoJ hike makes carry less profitable. On 31 July 2024 the BoJ raised to 0.25% → on 5 August the Nikkei fell 12.4% in a day, USD/JPY 161→141. Raising toward 1% could cut banks' current account balances at the BoJ from ¥454T to ¥280T and trigger $300–500 billion of outflows from Treasuries. Watch the BoJ 6–8 times a year.

Last time: August 2024

Then the Bank of Japan merely hinted at another hike. That was enough: the Nikkei fell 12.4% in one day, the dollar fell 11% against the yen in a week, and 10-year US yields jumped 0.2 percentage points over several days. It was a "mini crash" — just a hint of normalization, and the effect was already in the billions. Analogy: a neighbor lent you money at 0% for years — then his own problems appeared and the rate became 3%. When millions of such "neighbors" try to repay at once — that is a global "bank run."

Crutch #1. The endless T-bill machine

In 2023 the US government was stuck: Republicans in Congress refused to raise the debt ceiling, and default was weeks away. They struck a last-minute deal — on the condition that the Treasury "get creative" to avoid pressuring long rates.

Treasury Secretary Janet Yellen applied a trick described by economists Steven Miran and Nouriel Roubini back in 2022: instead of issuing long bonds (10–30 years), issue masses of short bills (T-bills, up to a year). The bills offered money-market funds a yield slightly above the Fed's reverse-repo rate (ON RRP) — and money from that liquidity "parking lot" flooded into T-bills. The Treasury got funded, and bank reserves barely moved.

One figureThe ON RRP "cushion" shrank from $2.25 trillion to almost zero in a year and a half (mid-2023 to end-2024). The Treasury drained the Fed's liquidity piggy bank.
The physics: how the Treasury outsmarted the FedBefore the move, funds hold $2.25T in ON RRP at 5.2%, and bank reserves are $3.2T. Yellen's move: T-bills yield 5.3% — 0.1% above RRP. Funds switch (nobody forces them — it is simply better). After the move, ON RRP = $0, the same $2.25T now sits in T-bills, and bank reserves are $3.25T (a ~5% change). Result: the Treasury is funded, the Fed formally lost nothing — but the cushion is gone, and the next T-bill will pull straight from bank reserves.

By Miran and Roubini's estimates, the maturity manipulation cut 10-year Treasury yields by about 0.25 percentage points — the equivalent of a 1% Fed rate cut. In other words, the Treasury "outbid" the regulator's own monetary policy. That used to be unthinkable.

Why it works (for now)

Short bills are ideal collateral for funds: short (low risk), dollar-denominated (zero currency risk), government-backed (zero credit risk). While T-bills sit with funds, they do not "press" on the long end of the curve — mortgage rates, auto loans, and corporate bonds stay lower than they would otherwise be. When you see "Treasury places $X billion of T-bills" in the news — that machine is running.

Analogy: you feed a cat allergic to cheap food. Every day you buy it expensive food with your own money, and the cat gets used to it. Money runs out — and you "feed the cat" with bills you issue yourself. It works as long as the cat keeps eating.

Crutch #2. The floor of bank reserves. Why the Fed no longer tightens

In 2008 the Fed's balance sheet was $900 billion, by 2014 — $4.5 trillion, by 2022 — $8.9 trillion. A printing press on full. In 2022 the Fed began "undoing" QE with its QT program: each month it stopped reinvesting interest, and money drained out of the system.

But each new T-bill issue (crutch #1) began pulling reserves straight out of banks. The ON RRP cushion that used to absorb that blow ran out at the end of 2024. After that — only falling bank reserves. And there the Fed hit its "lowest comfortable level of reserves" (LCLOR).

One figureBy 2025 the Fed had radically slowed QT: from $60bn/month in 2022 to $25bn/month in May 2024 and $5bn/month in April 2025. Officially — "reserves reached a comfortable level." Unofficially — beyond that point panic began.

What LCLOR means in plain words

Banks hold reserves at the Fed — the money they use to settle with each other daily. If there is too little, what happened in September 2019 occurs: overnight rates spike from 2% to 10%, the market "freezes," and the Fed injects hundreds of billions in an emergency. LCLOR is the level below which the system starts to "suffocate": roughly 8% of US GDP, or about $2.7 trillion.

Analogy: you have 100 rubles in the bank and a bill of 80 to pay. If the bill is 90 — tight, but you manage. If 110 — you don't, and the bank freezes the next operation. The Fed's "account" is the LCLOR limit. They have reached it, and the next dollar of deficit is a "bill of 110."

What this means in practiceThe Fed is formally "independent," but its balance sheet is effectively tied to Treasury issuance: as long as the Treasury prints T-bills, the Fed cannot shrink its balance sheet. This is a structural constraint, not politics — no new Fed chair can fix it. For you: more money stays in the system than in a normal scenario, so dollar savings lose purchasing power a bit faster — a few percent a year, which compounds visibly over a decade.

Crutch #3. The yen subsidy. While Japan keeps rates low, the world buys dollars

Remember the story from Part 1: since 1983 Japan has been pulled into the American financial system. Plaza 1985, the 1990 bubble, ZIRP, NIRP, YCC, a Bank of Japan balance sheet bigger than GDP. By 2024 Japan holds more than $1.1 trillion of US government debt — its largest foreign buyer.

Why? The yen is the "cheapest currency to borrow." You borrow yen at 0% (in 2016–2024 even at −0.1%), convert to dollars, buy Treasuries at 4–5% — the difference is pure profit. That is the carry trade. In 2023 alone its volume with the yen hit a record $1+ trillion. For the Treasury this is free demand for its debt: 10-year yields stay lower than they would be, and mortgages and autos are cheaper.

The physics: how carry works and where it breaksBorrow ¥15bn at 0.1% from the BoJ, convert at 150 ¥/$ into $100M, buy a T-bill at 4.3%. The spread is 4.2% = $4.2M a year "out of thin air"; with 10x leverage that is $42M of income on $10M of your own (420% a year on paper). Reversal: yen 150→135 per dollar (10% in a day, like 5 August 2024) turns a $100M debt into $111M — an $11M loss against $4.2M of annual income, three years of profit wiped out in one day. Margin call → selling Treasuries → yields rise → Nikkei falls → the yen strengthens further — a loop.
⚠️ Breaking pointEvery BoJ hike makes carry less profitable. On 31 July 2024 the BoJ raised to 0.25% → on 5 August the Nikkei fell 12.4% in a day, USD/JPY 161→141. Raising toward 1% could cut banks' current account balances at the BoJ from ¥454T to ¥280T and trigger $300–500 billion of outflows from Treasuries. Watch the BoJ 6–8 times a year.

Last time: August 2024

Then the Bank of Japan merely hinted at another hike. That was enough: the Nikkei fell 12.4% in one day, the dollar fell 11% against the yen in a week, and 10-year US yields jumped 0.2 percentage points over several days. It was a "mini crash" — just a hint of normalization, and the effect was already in the billions. Analogy: a neighbor lent you money at 0% for years — then his own problems appeared and the rate became 3%. When millions of such "neighbors" try to repay at once — that is a global "bank run."

Break scenarios: how this can fall apart

Three crutches — three ways any one of them can break first. None of the scenarios is "guaranteed," but all are structurally possible, and all have consequences for ordinary people, not just Wall Street.

Scenario A: the Bank of Japan normalizes its rate

What happens: the BoJ keeps raising in 2026–2027. The carry trade becomes unprofitable; funds close dollar positions and sell Treasuries to repay yen loans. In the US: 10-year yields jump 0.5–1.5 percentage points, and mortgage rates follow. For you: if you hold dollar assets or debt — sharp swings within hours; world markets fall 10–20% over a few weeks. This is a "repricing shock," not the end of the world.

Scenario B: the basis trade unwinds

What happens: in autumn 2026–2027 the market turns volatile, hedge funds with basis trades get margin calls and are forced to sell Treasuries — a cascade. In the US: the Treasury market (the most liquid in the world) seizes up for 1–3 days, and the Fed buys paper urgently, as in March 2020 — but the problem is bigger, because the basis trade has grown many times. For you: a short-lived shock that fades, but confidence in "risk-free" Treasuries is dented and pressure on LCLOR rises.

Scenario C: the US loses AAA

What happens: Fitch already cut the US to AA+ in 2023; if Moody's follows, some institutional buyers will be mandated to sell Treasuries. In the US: one more large buyer disappears; the Treasury must raise rates, the deficit widens — a vicious circle of "deficit → more debt → worse rating → fewer buyers → even higher rates." For you: the slowest but deepest scenario: mortgages above 8%, a volatile dollar, accelerating inflation.

What all three shareIn every case the regulator faces a choice: let the market find a new price (harsh repricing) or turn on the printing press directly. So far they have always chosen the latter. That is "quantitative fiscal dominance": the central bank loses the ability to shrink its balance sheet, and the only way out is to grow it further.

The digital Plan B. RLN: what regulators are building while we are not looking

If the system is too fragile to touch, what do regulators do? The answer: they build a replacement. Not for ordinary people, but between banks — and indirectly it will affect everyone.

RLN (Regulated Liabilities Network) puts three kinds of "digital money" on one platform:

  • wCBDC — wholesale central-bank digital money. Banks only.
  • Tokenized deposits — ordinary bank money as tokens: instant interbank transfers, no SWIFT.
  • Tokenized Treasuries — government bonds as digital tokens that can be settled instantly as collateral.

The core idea: in March 2020 a margin call was processed manually and took days — those days cost the world market a trillion dollars. RLN wants a margin call processed by a smart contract in milliseconds: a liquidation cascade can be stopped before it starts. But there is a catch: before, dealer banks «put out fires» manually — now an algorithm does it, and the regulator gets direct control over the whole system. That is both a "rescue from disaster" and a "new dependence" — on code that must never fail.

Who builds it: Project Agorá (BIS) — seven central banks, including the Fed, the Bank of England, and the Bank of Japan, plus HSBC, Citi, and JPMorgan. Prototypes already work; 2–4 years to production. For you this changes nothing yet, but on a 5–10 year horizon the questions of privacy and control over your money stop being science fiction.

For you personally: what this means and what to watch

This is not a "how to make money off the crisis" section, but "how not to be caught off guard." Four scenarios for an ordinary person:

  • Dollar savings: short-term swings are possible; in the long run purchasing power erodes slowly but steadily (2–4% a year from faster QE-driven inflation). This is not "the dollar dies tomorrow," but "it will be worth less in ten years."
  • A mortgage or loan in dollars: if 10-year yields break and hold above 6%, refinancing becomes expensive. While the window is open, think about locking in a low rate.
  • You live in a country whose currency depends on the dollar: any sharp move in USD/JPY, the euro, or Treasury yields echoes in your currency. The dollar is global — keep part of your savings in hard currencies or gold if you can.
  • You just live in 2026: don't panic — the world won't collapse tomorrow and won't go back to "the way it was." The next 3–5 years are a time of tectonic shifts, and understanding what exactly is shifting is already an advantage.
What to watch in 2026–2027Bank of Japan meetings (6–8 times a year): each hike risks thawing the carry trade; watch the Nikkei and USD/JPY for 24 hours. The 10-year US Treasury yield: a break above 5.5–6% is a signal that the market is starting to "scold" the government (currently about 4.2%). Moody's decisions: Fitch already cut; if Moody's follows — a trigger for institutional selling. T-bill issuance volumes: if the Treasury starts pushing on the long end, the short-bill trick stops working. SEC clearing deadlines: cash 31.12.2026, repo 30.06.2027; another delay means the basis trade is still too big to touch.

In short. The dollar rests on three temporary structures: endless T-bills from the Treasury, the "floor" of bank reserves from the Fed, and cheap yen from the Bank of Japan. Each was born as an emergency measure, and separately they do not work.

The regulators have a Plan B — digitalization via RLN, wCBDC, and tokenized Treasuries. The price is more control and less privacy. All the figures and charts are in the Appendix.