The Fed
& The
Treasury
The United States government can create dollars out of nothing. The United States government also has to borrow money, can run out of cash, and once nearly missed a payment because a legal ceiling stopped it from writing a check. Both of those sentences are true, and they are not a paradox. They are true of different institutions.
This course is about those two institutions — the Treasury and the Federal Reserve — and the narrow, load-bearing seam where they meet. Module 01 builds the map: what each one is, what each one can and cannot do, who appoints and fires and funds them, and how a single dollar physically moves between them.
One Government,
Two Money Machines
Start with the question that makes the whole subject click, because almost every confusion about the Fed comes from skipping it: why isn't there just one?
Imagine a country where the finance ministry both spends the money and prints it. There is a bill to pay and no tax revenue to pay it with, so the ministry prints the difference. Nothing stops this. There is no committee to argue with, no auction to clear, no interest rate to pay. It is the cheapest possible way to fund a government, and it is the reason the price level in such a country stops being an economic fact and becomes a political one — a lever the party in power can pull whenever paying a bill is more urgent than the value of everyone's savings.
Nearly every modern state has responded to that problem the same way: split the spender from the printer.
The Treasury
Can spend. Can borrow. Cannot create.
It collects taxes, writes every federal check, and sells bonds to cover the gap. It has a checking account with a balance, and that balance can hit zero. In cash terms it behaves like an enormous, very creditworthy household.
The Federal Reserve
Can create. Cannot spend.
It issues the dollars — both the paper notes and the electronic reserves banks hold — and it sets the interest rate at which they are lent. What it cannot do is appropriate money for a highway, a missile, or a benefit check. It buys and sells financial assets; it does not fund the government's choices.
That split is the design. But the two halves cannot be fully separated, because they share the same currency, and because one of them has to keep the other's cash somewhere. The Treasury banks at the Fed. Its checking account is a line item on the central bank's balance sheet. Every tax dollar that arrives and every federal dollar that leaves passes across that boundary.
So the interesting material is not "what does the Fed do" or "what does the Treasury do," taken separately. It is what happens on the seam — and that is where this module is heading. But first we need to know precisely what is on each side of it.
// The whole separation of powers, as a pair of functions. // Notice how much longer one of them has to be. function treasuryCanPay(amount, state) { if (amount > state.appropriated) return "NO — Congress never authorised this spending"; if (amount > state.cashInTGA) { // Not enough in the checking account, so it has to borrow first. if (state.debtOutstanding + amount > state.debtLimit) return "NO — the statutory debt ceiling blocks the borrowing"; auction(amount); // and it pays whatever yield the market demands } return "yes"; } function fedCanPay(amount) { // There is no check to write. The Fed pays by typing a larger number // into a bank's reserve account, and that number is not drawn down // from anywhere — the liability simply comes into existence. creditReserveAccount(amount); return "yes"; }
What Each One
Actually Is
"The Treasury" and "the Fed" get used as if they were two comparable boxes on an org chart. They are not remotely the same kind of object, and the difference is most of the story.
The Treasury: an ordinary cabinet department
The Department of the Treasury was created in 1789 and is structurally unremarkable. A Secretary runs it. The President nominates that Secretary, the Senate confirms them, and the President can fire them tomorrow morning for any reason at all — or no reason. There is no independence claim anywhere in the building.
Underneath the Secretary sit the bureaus that do the actual work, and it helps to know which one is which:
IRS
Collects the revenue. Money in.
Fiscal Service
The Bureau of the Fiscal Service pays the bills and issues the debt. It is the part of the government that literally sends the money — well over a billion payments a year — and the part that runs the borrowing.
BEP & Mint
The Bureau of Engraving and Printing prints the paper notes; the Mint strikes the coins. Manufacturing, not money creation — see the callout below.
Two hard constraints are worth memorising, because they are what makes the Treasury a borrower rather than a printer. It cannot spend a dollar Congress has not appropriated — the Constitution puts that in Article I — and it cannot borrow past a limit Congress sets. Both of its powers are on loan from the legislature.
"The Treasury prints the money." It prints the paper, which is not the same thing. Those notes say Federal Reserve Note across the top because they are a liability of the Federal Reserve, not of the Treasury. The BEP manufactures them and sells them to the Fed at roughly the cost of production; they only become money when the Fed issues them, and the Fed pays for them by debiting a bank's reserve account. Coins are the one genuine exception — they are a Treasury liability, and the Treasury books the profit on them. That profit is a rounding error next to the federal budget.
The Fed: a deliberately strange hybrid
The Federal Reserve was created in 1913, and it is not one organisation. It is three things bolted together, and the bolts are visible on purpose.
- 1The Board of Governors — a federal agency in Washington, D.C. Seven governors, appointed by the President and confirmed by the Senate, serving 14-year staggered terms so that one seat comes up every two years. The Chair and the Vice Chairs are governors designated to those roles, separately confirmed by the Senate for four-year terms in them. This part is unambiguously government.
- 2Twelve regional Reserve Banks — Boston, New York, Philadelphia, Cleveland, Richmond, Atlanta, Chicago, St. Louis, Minneapolis, Kansas City, Dallas, San Francisco. Each is a federally chartered corporation, not a federal agency. Their staff are not civil servants. Their presidents are not confirmed by the Senate — they are selected by the non-banker directors of their own board and approved by the Board of Governors in Washington. This part is unambiguously not ordinary government.
- 3The FOMC — the Federal Open Market Committee, which is where monetary policy is actually decided. Twelve votes: all seven governors, the president of the New York Fed permanently, and four of the remaining eleven presidents on an annual rotation. Seven of those twelve votes were confirmed by the Senate. Five of them never were. That is not an oversight; it is the compromise.
The hybrid shape is a fossil of the fight that produced it. In 1913 the banking interests wanted a private bankers' bank on the European model, and the populist wing wanted a government-run bank answerable to Congress. What passed was neither: twelve regional banks with private-sector features, supervised by a public board in Washington. Every argument about whether the Fed is "public" or "private" is really an argument about which half of that compromise you are looking at.
Yes, commercial banks hold stock in their regional Reserve Bank. No, that does not make the Fed privately owned in any sense you would recognise. Member banks are required by law to buy the stock; they cannot sell it, trade it, or borrow against it; it pays a dividend fixed by statute rather than by profits (6% for banks under $10 billion in assets, and for the large banks the lesser of 6% or the yield on the most recent 10-year Treasury auction); and it carries no claim on the Fed's earnings and no vote on interest rates. What it buys is the right to elect six of the nine directors of the district bank — and three of those six must, by law, not be bankers.
The honest summary: the stock is a compulsory membership fee with a coupon attached. But don't overcorrect either — the Reserve Banks really are not normal agencies, and their presidents really do vote on the country's interest rate without ever facing a confirmation hearing. Both halves are true, and the friction between them is a live political question, not a settled one.
The Control Map
"Who controls the Fed?" has no one-word answer, because control is split into four separate levers — who appoints, who can remove, who pays for it, and who it must answer to — and different people hold each one. Click a box to see all four for that actor.
Pick an actor
Who Controls Them
Now the prose version of that map, because the shape of the answer matters more than the individual facts.
The Treasury: a short chain
The President picks the Secretary and can remove them at will. Congress decides how much money exists to spend and how much may be borrowed. Courts decide whether a given action was legal. That is the whole chain, and it runs top to bottom without a break. When a President wants the Treasury to do something, the Treasury does it or the Secretary is replaced. This is why "Treasury policy" and "administration policy" are effectively synonyms, and why the Treasury Secretary is the government's economic spokesperson in a way the Fed Chair deliberately is not.
The Fed: four levers, four holders
The Fed's answer is layered, and the layers were built at different times for different reasons.
- 1Congress created it and can rewrite it. The Fed exists because of the Federal Reserve Act, a statute. Congress could amend the mandate, restructure the Reserve Banks, or abolish the whole thing with an ordinary law. This is the ultimate control, it is real, and it is almost never used — which is exactly why the Fed pays such close attention to Congress. Independence held at the pleasure of the legislature has to be re-earned continuously.
- 2Congress set the goals; the Fed picks the means. A 1977 amendment instructs the Fed to pursue "maximum employment, stable prices, and moderate long-term interest rates" — three things, universally called the dual mandate because the third is treated as a consequence of the first two. Note the division: the objectives are political and legislated. The instruments — which rate, how far, how fast — are the Fed's. Economists call this arrangement instrument independence without goal independence, and it is the modern template for central banks everywhere.
- 3The President appoints, but slowly. All seven governors are presidential appointments, and the choice of Chair is a genuinely powerful lever over the institution's direction. But the 14-year staggered terms mean a President inheriting a full Board fills roughly two seats per term in the normal course of events. The structure is designed so that no single administration can restaff monetary policy inside an election cycle.
- 4Removal is restricted — and this is the contested edge. The statute says a governor may be removed by the President "for cause." It does not define cause, and for most of the Fed's history nobody tested it: no President has successfully removed a Fed governor. Recent litigation over attempted removals has put real pressure on that clause, and the durability of the protection is now an open legal question rather than a settled assumption. Module 12 takes this up properly; for now, note only that the wall here is a phrase in a statute, not a constitutional guarantee.
- 5Nobody appropriates its budget. This is, structurally, the most important item on the list and the one people miss. The Fed receives no money from Congress. It earns interest on the trillions of securities it holds, pays its own operating costs out of that, and sends the remainder to the Treasury. An institution that does not have to ask for funding cannot be squeezed through the appropriations process — the standard tool for disciplining a federal agency simply does not have a handle to grab. Critics of Fed independence tend to start here, and they are pointing at the right thing.
- 6Accountability runs through disclosure rather than command. The Chair testifies before Congress twice a year on monetary policy. The FOMC publishes a statement the same day it decides, minutes three weeks later, and full meeting transcripts after five years. The Government Accountability Office audits most Fed activity, with monetary-policy deliberations carved out of that authority — the carve-out that "Audit the Fed" bills have repeatedly targeted. The financial statements are audited by an outside accounting firm. The theory is that an institution which must explain itself in public, in detail, on a schedule, is disciplined by that obligation even when nobody can overrule it.
Until 1935, the Secretary of the Treasury was a member of the Federal Reserve Board — its chairman, in fact — sitting alongside the Comptroller of the Currency. The government's chief borrower had a seat, and the gavel, at the table setting the price of borrowing. The Banking Act of 1935 removed both officials from the Board and created the FOMC in roughly its modern form. If you want a single date for when the Fed stopped being an arm of the Treasury and started being a counterparty to it, that is the one.
The phrase practitioners use for all of this is "independent within the government, not independent of it." It is not a slogan; it is a fairly precise description. Congress owns the goals and could dissolve the whole arrangement. The President owns the appointments. What the Fed owns is the decision itself, in the moment, plus a funding stream nobody has to vote on.
Where They Touch:
The TGA
Everything so far has been structure. Here is the mechanism, and it is one account.
The Treasury's checking account is called the Treasury General Account, and it is held at the Federal Reserve Bank of New York. Every dollar of federal revenue lands in it. Every federal payment — a tax refund, a Social Security deposit, a defence contractor's invoice, an interest coupon on a Treasury bond — leaves from it. The Fed is the Treasury's fiscal agent: it holds the account, clears the payments, and physically runs the debt auctions on the Treasury's behalf.
That relationship is easy to hear as a piece of trivia. It isn't. Follow the consequence:
The TGA sits on the liability side of the Fed's balance sheet — in the same column as bank reserves. There is a fixed amount of Fed liabilities in existence at any instant, and the TGA and reserves are both drawing from it. So when money moves into the government's account, it moves out of the banking system, and when the government spends, it moves back in. Not metaphorically. As a matter of double-entry bookkeeping.
// Everything on the Fed's liability side is somebody's claim on the Fed. // At any instant the total is whatever the asset side supports. const fedLiabilities = currency + reserves + tga + reverseRepo + capital; // Rearrange for the row the banking system actually cares about: const reserves = fedLiabilities - currency - tga - reverseRepo - capital; // Now read that as a sentence. Hold the asset side still — no QE, no QT, // no policy decision of any kind — and raise the TGA by 100 billion. // Reserves fall by exactly 100 billion. The Treasury just tightened // monetary conditions by paying attention to its own cash management.
This is the seam, and it means something uncomfortable for the clean separation we started with: fiscal actions have automatic monetary footprints. Nobody has to intend it. The Treasury does not decide to affect money markets when it collects corporate taxes in mid-April; it just collects them, and several hundred billion dollars of bank reserves vanish from the system in a few days as a mechanical side effect. The Fed then has to accommodate that if it wants its interest rate to stay where the FOMC put it.
Two labs, then. The first walks a single dollar all the way through, so the bookkeeping stops being abstract. The second lets you push the TGA around and watch what falls out.
Follow One Dollar
A toy economy: you, a contractor, one commercial bank, the Fed, the Treasury. Step through a tax payment, a government payment, and then a deficit financed by borrowing. Watch which rows move — and, more importantly, which totals don't.
The ledger
| Who holds it | What it is | Balance |
|---|
The ledger, as ~20 lines of code
Everything the step-through just did, written out. Change the numbers, add a call, break the invariant on purpose — it runs in this page, and nothing is saved.
The TGA Dial
Now the same identity at national scale. Hold everything else on the Fed's balance sheet still and split a fixed pool of Fed liabilities between two users: the government's checking account and the banking system's reserves. Drag the dial. The numbers here are illustrative round figures, but the seesaw is exact.
Treasury cash vs. bank reserves
Why this matters enough to be a whole lab: the size of these swings is not small. A corporate tax date, or the Treasury rebuilding its cash pile after a debt-ceiling standoff has forced it to run the account down, can move hundreds of billions of dollars of reserves in a matter of weeks. That is a monetary event of a size the FOMC would deliberate for months over — produced by nobody's monetary decision at all. It is the tax calendar and a statutory borrowing limit.
Which is the real reason the Fed and the Treasury talk constantly. The Treasury publishes its cash and borrowing projections in advance; the Fed's markets desk watches the account daily and adjusts its operations so the policy rate stays where the committee set it. The relationship is less "two rival powers" and more "two operators sharing one balance sheet, who need each other's calendar."
The Wall
Between Them
One question should be nagging by now. The Fed can create dollars. The Treasury needs dollars. They share an account. What actually stops the obvious shortcut?
A rule, and it is narrower than you would expect. The Federal Reserve may not buy Treasury securities directly from the Treasury. Section 14 of the Federal Reserve Act permits purchases of government debt only in the open market — from dealers, funds, and other investors who already own it. A limited direct-purchase authority did exist, capped at $5 billion outstanding, from 1942 onward; Congress let it lapse in 1981 and never restored it.
What the wall forbids
The Treasury cannot hand a freshly printed bond to the Fed and receive freshly created dollars in return. There is no overdraft facility. The government's account at the central bank cannot go negative.
What it permits
The Fed owning trillions of dollars of Treasury debt — which it does. It simply has to buy that debt second-hand, from whoever bought it at auction.
At which point the natural objection is: so what? If the Fed can buy an unlimited quantity of Treasury bonds the day after the auction, has the wall accomplished anything beyond adding a middleman?
Something, yes — and it is worth being precise about how much. What the rule guarantees is that every dollar the government borrows has to clear a real auction at a real price first. Investors show up, bid, and set a yield. If they are unenthusiastic, the yield rises and the government's borrowing costs rise with it, immediately and visibly. That price signal is the disciplining mechanism, and the wall is what keeps it from being bypassed. A finance ministry with a direct line to the printing press never has to find out what its debt is worth.
It is genuinely a thin wall. Central banks that buy heavily in the secondary market are pressing on it, and whether large-scale asset purchases amount to deficit financing by a slower route is one of the real arguments in monetary economics — Module 09 is about exactly that. But thin is not the same as absent. The distinction between the central bank owns a lot of government debt and the central bank is the government's overdraft is the distinction between a functioning monetary system and a currency crisis, and this clause is where it lives.
Clearing a Treasury auction
This is the mechanism the wall in 01.8 protects. The Fed cannot participate here — it can only buy from these bidders afterwards, once they have already set the price.
There is one narrow exception in practice: when Treasury securities the Fed already holds mature, the Fed rolls them over by exchanging them for new ones directly at auction. No new money is created and the Treasury raises no additional cash from the market — the Fed's position is simply extended. That is a bookkeeping continuation of an existing holding, not new financing, and it is the only routine direct transaction of its kind.
Recap &
Your Turn
Module 01 in seven lines:
- 1The spender and the printer are separated on purpose. The Treasury can spend and borrow but not create; the Fed can create but not spend.
- 2The Treasury is an ordinary cabinet department with a short, unbroken chain of command running up to the President and out to Congress's purse.
- 3The Fed is a hybrid: a federal board in Washington, twelve chartered corporations in the districts, and a twelve-vote committee that mixes Senate-confirmed governors with presidents who never faced a hearing.
- 4Control over the Fed is split four ways — Congress appoints the goals and could repeal the whole statute, the President appoints the people, removal is limited by a contested "for cause" clause, and nobody at all appropriates the budget.
- 5The two institutions meet at one account: the Treasury General Account, held at the New York Fed, sitting on the Fed's liability side next to bank reserves.
- 6Because of that shared column, taxing and spending mechanically drain and add bank reserves. Fiscal actions carry monetary side effects nobody chose.
- 7The wall is that the Fed may buy Treasury debt only in the open market. Every borrowed dollar must clear a real auction first.
Answer before you peek
Question: Of everything that protects the Fed's independence — the 14-year terms, the "for cause" removal clause, the split appointment structure — which single feature would you remove if you wanted to bring the Fed under political control most effectively, and why?
Test Yourself
Five questions on Module 01. Pick an answer for each, then submit.
Inside The Treasury:
The Checkbook
There is a version of the Treasury that exists in political argument: an institution that decides how much the country spends. That institution does not exist. Congress decides what is spent; the Treasury finds out, and then has to make the payment clear.
What the Treasury actually is, operationally, is the largest cash-management operation on earth — well over a billion payments a year, a checking account that swings by hundreds of billions in a week, and a borrowing programme that has to raise money on a fixed schedule regardless of whether the timing is convenient. This module is about how that machine works, because almost every conflict with the Fed in later modules comes out of its mechanics rather than anyone’s intentions.
The Cashier’s Problem
Start with the constraint that shapes everything else. Federal money comes in lumpy and goes out smooth.
Outflows are relentless and largely fixed: Social Security on a published schedule, salaries, Medicare reimbursements, interest coupons, contractor invoices. Inflows are violently uneven. Corporate estimated taxes arrive on four days a year. Individual filing season dumps an enormous receipt in mid-April and a large refund outflow in February and March. On a single mid-April day the account can take in more than it does in an ordinary fortnight.
So the Treasury faces the problem any cashier faces, scaled to a national budget: it must be able to pay everything it owes on the day it owes it, out of an account whose inflows do not line up with its outflows. It has exactly three instruments for that — the cash it is holding, the taxes arriving, and the debt it can sell — and the third one has a legal ceiling on it.
Every later fight in this course — the debt ceiling, the reserve drains, the coordination calls with the Fed’s markets desk — is a downstream consequence of this one mismatch. The Treasury is not trying to influence monetary conditions. It is trying to not bounce a cheque, and monetary conditions move as a side effect.
Who Actually
Moves The Money
The Secretary does none of this. Two bureaus do nearly all of the work, and they were one organisation only recently.
Bureau of the Fiscal Service
Formed in 2012 by merging the Financial Management Service (which paid the bills) with the Bureau of the Public Debt (which borrowed the money). Putting them together was an admission that they were always the same job: the payments schedule and the borrowing schedule are the two halves of one cash forecast.
It disburses the overwhelming majority of federal payments and runs every auction.
Internal Revenue Service
The inflow side. Worth noting that the IRS does not hold the money it collects for even a moment longer than it takes to settle — receipts land in the Treasury General Account at the Fed, which is the only account that matters for anything in this course.
Alongside them sit the manufacturing bureaus from Module 01 — the Bureau of Engraving and Printing and the Mint — plus the policy offices (domestic finance, international affairs, OFAC and the sanctions apparatus) that give the department its public voice. None of those touch the plumbing.
The plumbing itself is the interesting part, because the Treasury does not own it.
How A Payment
Leaves The Building
Follow one Social Security deposit. The striking thing is how little of it the Treasury does itself.
Notice the third step, the self-message. That is the same double-entry move from Module 01 — TGA down, reserves up — and it is happening here at a scale of tens of billions in a morning. Social Security payment days are visible in the reserve data. Nobody at the Fed decided that reserves should rise on the third of the month; the payment calendar decided it.
The reason the Treasury uses the Fed rather than a commercial bank is partly historical and partly structural. Until 2008 the Treasury did keep large balances at commercial banks, in Treasury Tax and Loan accounts, specifically so that tax receipts would not drain reserves the moment they were collected. That arrangement was largely wound down during the crisis, and the cash now sits at the Fed. One consequence: the reserve swings this course keeps pointing at got considerably larger, as a side effect of an operational cleanup.
The Daily
Treasury Statement
Every business day the Treasury publishes exactly what moved: opening balance, deposits, withdrawals, debt issued, debt redeemed, closing balance. It is the most boring document in American finance and one of the most closely read — money-market desks pull it the moment it lands, because it tells them what happened to reserves.
Build a day. Then build a bad one.
Close the books on one day
Two things to try. Set new debt to zero and push withdrawals well above deposits — that is a debt-ceiling day, and you can watch the buffer disappear at a rate that tells you exactly how many days of room the government has left. Then set deposits to $300bn with everything else normal: that is corporate tax day, and the closing balance jumps by more than most countries hold in reserves.
Why Hold
Cash At All?
Here is a question worth sitting with before reading the answer. The Treasury can sell debt almost any day it likes, at enormous size, into the deepest securities market in the world. So why hold a cash buffer at all? Every dollar sitting idle in the TGA is a dollar the government borrowed and is paying interest on while it does nothing.
The answer is that “can sell debt any day it likes” turned out to be an assumption rather than a fact, and it failed twice within five years.
- 12008. Funding markets seized. The premise that a sovereign borrower always has access stopped being something anyone wanted to rely on without a fallback.
- 2Hurricane Sandy, October 2012. A storm shut down the market infrastructure in lower Manhattan. Not a credit event, not a loss of confidence — weather. The government could not auction for days, and the only thing standing between it and a missed payment was cash already in the account.
So in 2015 the Treasury adopted an explicit policy: hold enough cash to cover roughly one week of outflows even if it cannot borrow at all, subject to a floor. In the years since, that target has generally run in the high hundreds of billions of dollars.
That policy is a purely prudential decision about operational risk — the fiscal equivalent of keeping an emergency fund. But because the buffer lives in the TGA, and the TGA is a Fed liability, choosing to hold several hundred billion dollars of insurance permanently removes several hundred billion dollars of reserves from the banking system. A risk decision at one institution became a standing monetary fact at the other, and nobody had to intend it.
Regular And
Predictable
A corporate treasurer times the market. They watch rates, and when borrowing looks cheap they issue; when it looks expensive they wait. This is completely normal behaviour and the US Treasury refuses to do it.
The doctrine is called regular and predictable, and it is the closest thing the department has to a philosophy. The Treasury announces a calendar of auctions — which securities, roughly what size, on which dates — and then executes it more or less regardless of where yields are. Quarterly refunding statements in February, May, August and November tell the market what the next three months will look like, informed by an advisory committee of market participants.
Why give up the chance to time the market? Because the Treasury is not a participant in this market, it is the market, and its incentives are inverted:
It cannot stop
A company can decline to borrow this quarter. The government must fund the deficit and roll the maturing stock no matter the conditions. An issuer who cannot walk away has no bargaining position to preserve.
Surprise costs money
Buyers price uncertainty. If nobody knows what size is coming, every bidder builds a cushion into their bid — and the Treasury pays that cushion on every dollar, forever.
The horizon is infinite
Clever timing might save a few basis points once. Being the most boring, most liquid borrower in the world lowers the cost of every auction for decades. That trade is not close.
The exception proves the rule: cash management bills, irregular short-dated issues used to plug a genuine timing gap. They exist precisely so that the regular calendar never has to be disturbed.
// What the Treasury has to raise in a quarter is NOT the deficit. // It is three things, and only the first is what the news reports. function financingNeed(q) { return q.deficit // the genuine shortfall + q.debtMaturingThisQuarter // the treadmill — usually the big one + (q.targetCashBalance - q.cashOnHand); // refilling the TGA } // Which is why a debt-ceiling standoff is so violent when it ends: // the third term, normally near zero, suddenly becomes hundreds of // billions, and every one of those dollars is drained from bank reserves. const q = { deficit: 500, // $bn, one quarter debtMaturingThisQuarter: 6200, targetCashBalance: 850, cashOnHand: 60 // ran it down during the standoff }; financingNeed(q); // => 7490 — of which only 500 is new borrowing
Gross Issuance
Is Not Borrowing
Which sets up the number that misleads more people than any other in federal finance.
The reconciliation is the code block above: what the Treasury raises is the deficit plus everything maturing plus any change in the cash buffer. In a normal year the middle term dwarfs the other two, because short-dated bills mature and are reissued constantly. A four-week bill rolled all year is thirteen auctions and zero new debt.
This also explains why the composition of the debt is a live policy question rather than an accounting detail. If the whole stack were thirty-year bonds, the treadmill would nearly stop. If it were all four-week bills, the entire national debt would reprice every month. Where the Treasury sits between those two poles determines how fast a change in interest rates reaches the budget — which is the next lab.
The Maturity Dial
Drag the average maturity of the debt and watch two numbers fight each other: what the government pays each year, and how fast a rate shock reaches it.
The maturity dial
A stylised model on $28 trillion of debt held by the public, with a gently upward-sloping curve. The numbers are approximations; the trade-off they describe is exact.
Short is cheap until it isn’t. Financing the government with four-week bills usually costs less, because short rates are usually lower than long ones. But it means refinancing the entire national debt roughly every month, at whatever rate exists that month — so a rate shock reaches the whole stack almost immediately. Financing long costs more every single year and buys insurance: a shock takes a decade to work through. The Treasury’s actual answer has been to sit near the middle and, above all, not to move around much.
The Two Debts
One last piece of Treasury anatomy, because it is the source of a genuinely confusing pair of headline numbers. There are two debt figures, they differ by trillions, and both are correct.
Debt held by the public
Securities owned by people who are not the federal government: households, pension funds, foreign central banks, banks — and the Federal Reserve, which counts as public for this purpose. This is the number that was actually borrowed in a market, and the one that matters for interest costs and for everything in this course.
Intragovernmental holdings
Securities the government owes itself. When the Social Security trust fund runs a surplus, it is required by law to hand the cash to the Treasury and take non-marketable bonds in return. Real legal obligations, but no auction happened and no outside investor was involved.
Add them and you get the number that shows up on debt clocks. Neither figure is fake, but they answer different questions: one asks what the government owes, the other asks what the government had to borrow from someone else. Only the second one has a market price, a yield, and a bidder who can say no.
Recap &
Your Turn
Module 02 in six lines:
- 1Federal money arrives lumpy and leaves smooth. Everything the Treasury does operationally follows from that mismatch.
- 2The Bureau of the Fiscal Service pays the bills and sells the debt — one bureau, because they are one cash forecast.
- 3The Treasury certifies payments; the Fed moves them, over ACH for bulk and Fedwire for urgent. Payment days show up in reserve data.
- 4The cash buffer exists because market access failed in 2008 and again in a 2012 storm. Since 2015 the target is about a week of outflows.
- 5Issuance is regular and predictable by doctrine: an issuer who can never walk away buys cheaper debt by being boring than by being clever.
- 6Financing need = deficit + maturing debt + change in the cash balance. Gross issuance is mostly the middle term, and it is a treadmill, not a debt.
Answer before you peek
Question: The Treasury could cut its interest bill right now by financing everything with four-week bills, since short rates are usually below long rates. Give the strongest argument against doing it — and then give the strongest argument that the objection is overstated.
Test Yourself
Five on Module 02.
Inside The Fed:
The Odd Shape
Module 02 went inside the Treasury and found a cashier. This one goes inside the Federal Reserve and finds something stranger: a federal agency, twelve corporations, a payments utility, a bank supervisor and a research institute, all wearing one name.
The strangeness is not decoration. Every part of it is a fossil of an argument someone won in 1913, 1935 or 2010, and knowing which argument explains behaviour that otherwise looks arbitrary.
What The Board
Does All Day
Seven governors, one building on Constitution Avenue, and a job description that is mostly not monetary policy.
Supervision & regulation
The Board writes the rules for bank holding companies and state member banks, and runs the annual stress tests. Since 2010 there has been a dedicated Vice Chair for Supervision — a Senate-confirmed post created because Congress concluded that before the crisis, nobody in particular had owned this.
The plumbing
The Board oversees the payment system the Reserve Banks operate, and sets the terms on which banks get emergency liquidity. Reserve requirements are also its call — currently set at zero, which is itself a policy choice we come back to in Module 07.
Governing the twelve
It approves every Reserve Bank president, every Reserve Bank budget, and every change to a district's discount rate. This is the lever that makes twelve independent corporations behave as one system.
Consumer protection, partly
Historically a large part of the job. Dodd-Frank moved most rule-writing to the new CFPB in 2010, leaving the Fed with examination duties for the banks it supervises.
Notice what is missing from that list: the Board does not trade, does not hold your bank's money, and does not clear a single payment. All of that happens somewhere else.
What A Reserve Bank
Does All Day
A Reserve Bank is an operating company. Strip away the mystique and it runs four businesses.
- 1It is a bank for banks. Every reserve account lives at a Reserve Bank, and so does the Treasury General Account. When Module 01 said money “left the banking system,” this is the ledger it moved to.
- 2It is the emergency lender. The discount window sits here — a bank short of cash pledges collateral to its district bank, not to Washington. The rate is set by the district's own board and approved by the Board of Governors, which is why you occasionally see a district vote for a different rate than the one that ends up applying.
- 3It is a payments utility. Fedwire for large-value transfers, FedACH for bulk, FedNow for instant retail settlement since 2023, plus the physical business of shipping, sorting and destroying currency. This is genuinely enormous and genuinely boring.
- 4It is a research department with a listening post. Each bank employs economists and, more usefully, has directors and business contacts across its district. The Beige Book — published eight times a year, two weeks before each FOMC meeting — is that intelligence written down.
In 1913 the objection to a single central bank was not economic, it was geographic: a bank run from New York would see the country through New York's eyes. Twelve districts with local boards was the answer. Whether it still makes sense is a fair question — Missouri has two Reserve Banks and the entire western third of the country shares one, because the map was drawn around where the economic weight sat in 1914. But the listening function is real, and it is the reason a Dallas or Minneapolis president can walk into a meeting with something the Washington staff genuinely does not have.
Three Classes
Of Director
Here is the piece of the design that carries the most weight and gets the least attention: who sits on a Reserve Bank's board, and what each of them is allowed to decide.
Nine directors, in three classes of three.
Class A — the bankers
Elected by the member banks in the district, and they are bankers. This is the private-sector residue of the 1913 compromise, right out in the open.
Class B — elected, but not bankers
Also elected by the member banks, but they may not be an officer, director or employee of one. They are supposed to represent the district's broader economy — manufacturing, agriculture, labour, services.
Class C — appointed from Washington
Appointed by the Board of Governors, and held to a stricter bar: no bank employment and no bank stock. The Board also picks which Class C director chairs the whole board.
Now the rule that matters. Only Class B and Class C directors vote on who becomes president of the Reserve Bank — and even then, the Board of Governors in Washington has to approve the choice.
That was not always true. Before the Dodd-Frank Act in 2010, Class A directors — the bankers — voted too, which meant the people being supervised had a formal say in choosing their supervisor. Congress removed it in one line. It is the single cleanest example of how this institution actually changes: not by grand redesign, but by a clause.
// Dodd-Frank, 2010: the bankers stopped getting a say in who runs // the bank that supervises them. One line of the statute, and this // is the whole of it. function selectPresident(directors, candidate) { const voters = directors.filter(d => d.class === "B" || d.class === "C"); // ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ // Class A directors ARE bankers, elected by the member banks. // They sit on the board, they just cannot vote on this. const yes = voters.filter(v => v.votesFor === candidate).length; if (yes * 2 <= voters.length) return "not selected"; // And even a clean win is only a nomination. return boardOfGovernors.approves(candidate) ? "president, for a five-year term" : "blocked in Washington"; }
Choose A President
The rule is easy to state and easy to get wrong. Try it.
Hold the vote
Nine directors sit on a Reserve Bank board. Click the ones you want backing your candidate, then hold the vote.
New York
Is Different
One of the twelve is not like the others, and the reasons are historical rather than legal.
The New York Fed runs the Desk. When the FOMC votes, it issues a directive, and a trading floor in Manhattan executes it — every open-market purchase and sale for the entire System, managing a portfolio measured in trillions. The other eleven Reserve Banks own shares of that portfolio; only New York touches it.
It also conducts foreign-exchange operations, both for the Fed and as agent for the Treasury's Exchange Stabilization Fund — a dual-hat arrangement we come back to in Module 11 — and holds custody accounts for foreign central banks, which is why a meaningful share of the world's official gold sits in a vault under Liberty Street.
And its president is the only one with a permanent FOMC vote, plus the traditional role of the committee's Vice Chair. Module 04 is about the room they sit in.
The Reserve Bank closest to Wall Street is also the one that trades with Wall Street, supervises the largest institutions on Wall Street, and holds a permanent vote. Every reform proposal aimed at the Fed's structure eventually arrives at New York, and every one of them runs into the same practical objection: the Desk has to be somewhere, and it has to be where the market is.
The Whole Shape,
On One Page
Everything above, drawn once.
Recap &
Your Turn
Module 03 in 6 lines:
- 1The rate decision is eight days a year; supervision, payments and the discount window are the other three hundred and fifty-seven.
- 2The Board writes rules, runs stress tests, and approves every Reserve Bank president, budget and discount rate. It operates nothing.
- 3A Reserve Bank is an operating company: bank for banks, emergency lender, payments utility, and regional listening post.
- 4Nine directors in three classes. Class A are bankers; B and C are not; only B and C vote for the president, and Washington must approve.
- 5That voting rule was rewritten by Dodd-Frank in 2010 — before then, the supervised helped choose their supervisor.
- 6New York runs the Desk for the whole System, does the FX operations, and holds the only permanent FOMC vote.
Answer before you peek
Question: The twelve Reserve Banks are technically owned by the commercial banks in their districts, and three of every nine directors are bankers. Why is it still wrong to call the Fed “captured” on the strength of those two facts — and what is the strongest version of the worry that remains?
Test Yourself
Five questions.
The FOMC:
Twelve Votes
Module 03 drew the institution. This is the one room in it where the two halves of the 1913 compromise sit down together and decide something.
It is worth being precise about the mechanics here, because almost every popular account of “the Fed raised rates” compresses a nineteen-person deliberation, a twelve-person vote, and a trading desk in Manhattan into a single actor with a single will. None of those three things is the same as the others.
Twelve Votes,
Nineteen People
Nineteen people sit around the table. Twelve of them can vote. Both numbers matter, and the gap between them is the most misunderstood thing about this committee.
The twelve are the seven governors, the president of the New York Fed, and four of the remaining eleven presidents. Those four seats rotate, and the rotation is not discretionary — it is written into the Federal Reserve Act as four groups on fixed cycles.
Group 1 — one seat
Boston, Philadelphia, Richmond. Three banks, three-year cycle.
Group 2 — one seat
Cleveland and Chicago. Two banks, so they simply alternate — each votes every other year.
Group 3 — one seat
Atlanta, St. Louis, Dallas.
Group 4 — one seat
Minneapolis, Kansas City, San Francisco.
That is the whole rule, and it fits in nine lines of code.
// Twelve votes: seven governors, New York, and four rotating seats. // The rotation is not a lottery — it is a fixed cycle in statute. const NEW_YORK = "New York"; // votes every single year const GROUPS = [ ["Philadelphia", "Richmond", "Boston" ], // 3-year cycle ["Chicago", "Cleveland" ], // 2-year cycle ["Dallas", "Atlanta", "St. Louis" ], // 3-year cycle ["Minneapolis", "San Francisco","Kansas City"] // 3-year cycle ]; function votingPresidents(year) { const n = year - 2023; // any anchor year works return [NEW_YORK, ...GROUPS.map(g => g[((n % g.length) + g.length) % g.length])]; } votingPresidents(2026); // => ["New York", "Philadelphia", "Cleveland", "Dallas", "Minneapolis"] // The other seven presidents still attend, still argue, still publish a dot. // They simply do not vote that year. Nineteen in the room, twelve on the roll.
A president without a vote this year is not a spectator. They attend every meeting, receive every briefing, speak in both go-rounds with the same standing as a governor, and publish a dot in the projections. What they cannot do is be counted on the roll. So when a non-voting president gives a speech disagreeing with the decision, that is not a rebel outside the tent — it is a member of the committee who argued and lost, and who will be voting again in a year or two.
Who Votes This Year?
Drag the year. The four rotating seats move on cycles that have run without interruption for decades — which means you can read them forwards or backwards.
Who votes this year?
Two Days
In The Room
The eight meetings a year run two days, and the shape of them is almost ritualised.
It begins before anyone arrives: the staff circulate briefing books days in advance — the forecast, the policy options, the scenarios. Then two go-rounds. In the economic go-round, every one of the nineteen speaks about conditions, national and in their own district. In the policy go-round, every one of them says what they think should be done.
Only then does the Chair propose something — and the Chair has just heard, out loud and in order, where the room already is. The vote follows, and the directive goes to the Desk in New York, which has to go and make the rate actually be that. Module 07 is about how.
The Dot Plot,
And What It Isn't
Four times a year the committee publishes the Summary of Economic Projections, and buried in it is the chart that moves markets more than the statement does: the dot plot.
Each of the nineteen participants marks where they think the policy rate should sit at the end of this year, the next few years, and in the longer run. The dots are anonymous. They are not a forecast of what the committee will do, they are not negotiated, and — the part everyone forgets — they are not a commitment to anything. Each dot is one person's view of the appropriate path conditional on their own economic forecast. If the economy does something else, the dot was never a promise.
Markets read the median anyway, because the median is the closest thing to a number the committee will give them. Which creates a specific fragility worth seeing for yourself.
Move The Dots
Move the dots
Nineteen participants each mark where they think the policy rate should end the year. Slide people from the lower path to the higher one and watch what the median does — and what it refuses to do.
Two lessons fall out. First, the median is a step function — it sits still while opinion shifts underneath it, then jumps by a quarter point when the tenth person crosses over. Markets that trade the median therefore see nothing, nothing, nothing, lurch.
Second, the median can be a position nobody actually holds as their preferred path, and it tells you nothing about how strongly it is held. A 10–9 split and a 19–0 consensus can print the same dot.
Dissent, And Why
It Is Rare
Most FOMC votes are unanimous or near it, and people read that as groupthink. It is mostly structural.
By the time a vote is called, the Chair has proposed the option the policy go-round already pointed at. A Chair who repeatedly proposes something the room does not support stops being able to lead it — and the room refreshes every year as the rotation turns. So the negotiation happens in the go-round, and the vote records the result.
Which is what makes a dissent informative. A dissenting governor or president is saying they could not live with the compromise, in public, on the record, with their name attached. Historically dissents cluster at turning points — when the committee is late to tighten, or arguing about when to stop. One dissent is noise. Three in the same direction, twice running, is the committee telling you it is about to change its mind.
The statement lands at 2pm the day it is decided; the changed words matter more than the kept ones. The press conference follows half an hour later and is where the Chair adds the qualifications the statement could not carry. The minutes arrive three weeks later with the shape of the disagreement — “several participants” and “a few participants” are terms of art. The full transcripts come out after five years, by which point they are history rather than news, which is exactly the point of the delay: people speak more freely when the verbatim record is a long way off.
Recap &
Your Turn
Module 04 in 6 lines:
- 1Nineteen participants, twelve votes: seven governors, New York permanently, and four rotating seats on fixed statutory cycles.
- 2Non-voting presidents attend, argue and publish dots with the same standing. Only the roll call excludes them.
- 3Two days: staff briefings, an economic go-round, a policy go-round, then the Chair proposes what the room already pointed at.
- 4The directive goes to the Desk in New York, which has to go and make the rate actually be that.
- 5The dot plot is nineteen anonymous conditional views, not a forecast and not a promise — and its median is a step function.
- 6Unanimity is mostly structural, which is what makes a dissent worth reading. Clusters of them mark turning points.
Answer before you peek
Question: The Chair has one vote out of twelve and cannot fire anyone on the committee. Where does the Chair's power actually come from — and what is the check on it?
Test Yourself
Five questions.
Reserves & The
Balance Sheet
This is the most technical module in the course and the one that pays for itself fastest. Everything after it — the floor system, remittances, QE, the debt ceiling — is an argument about lines on one balance sheet.
The good news is that the balance sheet is genuinely simple. It is the intuitions people bring to it that are complicated, and most of this module is about dismantling them.
What A Reserve
Actually Is
We have been saying “reserves” for four modules. Time to be exact, because almost everything that follows depends on it.
A reserve is a deposit a bank holds at the Federal Reserve. That is all. It is an asset to the bank and a liability of the Fed, exactly the way your deposit is an asset to you and a liability of your bank. Reserves are the settlement asset of the American financial system: when one bank owes another, the debt is finally extinguished by moving reserves.
Who can hold them
Banks, and a short list of other institutions with Fed accounts. Not you, not a hedge fund, not a corporation. This exclusivity is why the reserve market and the deposit market are two different markets with two different interest rates.
What they are for
Settling payments, and satisfying whatever the Fed requires banks to hold. On that second point: reserve requirements have been zero since March 2020. The Fed eliminated them, and almost nothing happened — which tells you they had stopped being the binding constraint long before.
What they are not
Not vault cash, not a rainy-day fund the bank chooses to keep, and emphatically not money that is “sitting idle instead of being lent out.” That last one is the most common error in popular commentary, and 05.4 is about why.
Where Reserves
Come From
Not from taxes, not from deposits, not from anywhere. Watch the second line.
// This is the entire mystery of money creation, and it is four lines. function fedBuysSecurity(priceInDollars, sellerBank) { fed.assets.securities += priceInDollars; // the Fed now owns a bond fed.liabilities.reserves += priceInDollars; // ...and owes this, to a bank sellerBank.assets.reserves += priceInDollars; // which the bank now holds return "done"; } // Read the second line again. The reserves were not moved from another // account. There was no account. The Fed wrote a number into its own // liability column, and that number is now money the banking system holds. // // This is not a trick and it is not unique to central banks — every bank // creates a deposit the same way when it makes a loan. What IS unique is // that the Fed's liabilities are the settlement asset for everyone else, // so nobody can refuse them, and nobody can redeem them for anything but // currency, which is another Fed liability. function fedLetsItMature(amount, holderBank) { fed.assets.securities -= amount; fed.liabilities.reserves -= amount; // and now they are gone holderBank.assets.reserves -= amount; } // That second function is quantitative tightening. Module 09.
It is worth sitting with how ordinary this is. The Fed buys an asset and pays for it by crediting an account it maintains itself. The constraint on doing this is not the mechanics — there is no mechanical limit — it is that the Fed has a legal mandate and an inflation target, and buying assets with created reserves affects both. The limit is a policy limit, not an accounting one, and confusing the two is the source of a lot of bad argument in both directions.
The Balance Sheet,
With Handles
Reserves are the residual. Set everything else and see what is left over for the banking system.
Set the balance sheet, read the reserves
Only the top slider is a monetary policy decision. The other three are “autonomous factors” — the public deciding how much cash to carry, the Treasury managing its own account, money funds parking overnight. Each one competes with the banking system for the same pool of Fed liabilities, and none of them was voted on by anybody.
The Hot Potato
Here is the fact that breaks most people's intuition, and it is worth slowing down for.
An individual bank can get rid of reserves. The banking system cannot.
Take a bank with reserves it does not want. It can lend them overnight to another bank — the reserves move, and now the other bank has them. It can buy a security — the seller deposits the proceeds somewhere, and that bank's reserve account rises. It can make a loan to a customer, who spends the money, and it lands as a deposit at some other bank, whose reserves go up by exactly the amount that left.
At every step the reserves are still there. They have simply changed which bank is holding them. Economists call this the hot potato: the aggregate quantity is fixed by the size and composition of the Fed's balance sheet, and no amount of banking activity changes it. Only the Fed can create or destroy reserves.
“Banks are sitting on trillions in reserves instead of lending it out” is not a criticism of banks — it is a category error. Lending does not reduce reserves; it moves them. The system holds whatever the Fed's balance sheet obliges it to hold, and the only way for the total to fall is for the Fed to shrink its assets or for one of the competing liabilities — currency, the TGA, reverse repo — to grow. This is also why the old textbook “money multiplier,” in which reserves are lent out and re-deposited several times over, does not describe the modern system.
Scarce, Ample,
Abundant
Which sets up the question Module 07 answers in full: how much is enough?
Before 2008 the Fed deliberately kept reserves scarce — a few tens of billions, system-wide. That scarcity was the whole mechanism: because banks genuinely needed reserves and there were barely any, small open-market operations could move the overnight rate a long way. The Desk added or drained a few billion and the fed funds rate went where the FOMC wanted.
After 2008 the balance sheet grew by trillions, and scarcity ended. In an ample-reserves regime the old lever stops working — adding ten billion to a system holding three trillion does nothing at all — so the Fed had to find a different way to set the rate. That is the floor system, and it is Module 07.
One episode is worth knowing now, because it shows the boundary is real rather than theoretical. In September 2019, after a couple of years of balance-sheet runoff colliding with a corporate tax date and a large Treasury settlement, overnight repo rates spiked violently. Reserves had quietly become scarce again and nobody had known exactly where the line was. The Fed added reserves and the spike passed — but the lesson stuck: “ample” is not a number anyone can compute in advance, and you find the floor by hitting it.
Recap &
Your Turn
Module 05 in 6 lines:
- 1A reserve is a bank's deposit at the Fed — an asset to the bank, a liability of the Fed, and the settlement asset for the whole system.
- 2Only banks and a few other account holders can hold them, which is why the reserve market and the deposit market price differently.
- 3Reserve requirements have been zero since March 2020, and almost nothing happened.
- 4Reserves are created when the Fed acquires an asset and destroyed when it lets one run off. Nothing else creates or destroys them.
- 5An individual bank can shed reserves; the system cannot. Lending moves them, it does not consume them — the money multiplier does not describe this.
- 6Scarcity was the pre-2008 mechanism and it is gone. September 2019 proved the boundary is real and cannot be computed in advance.
Answer before you peek
Question: If lending does not consume reserves, and reserve requirements are zero, what actually stops a bank from making unlimited loans?
Test Yourself
Five questions.
Birth Of A
Treasury Bond
Module 01 established the wall: the Fed may only buy Treasury debt second-hand, so every borrowed dollar has to clear a real auction at a real price first. This module is that auction, in detail.
It rewards the attention. The Treasury auction is one of the most carefully engineered markets in the world — every feature of it, down to the pricing rule, exists because some earlier version cost the taxpayer money.
The Announcement
A Treasury bond does not appear. It is announced, priced by a market that does not yet own it, auctioned on a fixed date, and settled days later. Four steps, and each one exists to solve a specific problem.
The announcement comes days ahead: how much, what term, the identifying CUSIP, the auction date, the settlement date. This is Module 02's regular-and-predictable doctrine made concrete — nobody is surprised, because surprise is expensive.
What happens next is the part that people find genuinely odd. Between announcement and auction, the security trades. It does not exist, nobody owns it, and there is an active market in it anyway.
A Price Before
The Bond Exists
This is the when-issued market: a forward market in a bond that will exist on settlement day. Dealers buy and sell it, take positions in it, and hedge with it, all before a single unit has been created.
It sounds like a curiosity. It is actually the most important price in the whole process, because it is what makes an auction gradeable.
Think about what you would otherwise know after an auction. The government sold $50 billion at 4.29%. Is that good? Bad? Compared to what? Without a market price established beforehand, the only answer is “compared to nothing.” The when-issued yield at the bidding deadline is the market's honest estimate of where the bond should clear — so the gap between it and the actual result is a clean measure of whether demand showed up.
That gap has a name. Tail = stop-out yield − when-issued yield.
A positive tail means the Treasury had to go further up the ladder than the market expected — it paid more, demand disappointed. A negative tail means the auction “stopped through”: it cleared at a lower yield than expected, and demand was strong. Tails are measured in tenths of a basis point and traders care about them intensely, which sounds absurd until you remember the sizes involved.
Auction Day
Announcement to allotment, in order.
Two Kinds Of Bid,
One Price
Two kinds of bid go into an auction, and they behave completely differently.
Competitive
You state a yield you are willing to accept. If the auction clears at or above it, you are filled; if it clears below it, you get nothing. No single bidder may take more than 35% of the offering. This is where the price is discovered, and it is overwhelmingly institutional.
Noncompetitive
You state a quantity and accept whatever yield the auction produces. Capped at a few million dollars per bidder — small enough that this is the retail door, and it is how an individual buys directly. Noncompetitive bids are filled first, off the top, before the competitive ladder is walked at all.
Then the mechanism from Module 01's sandbox: the Treasury sorts competitive bids from lowest yield to highest and fills them in order until the size is covered. The last yield accepted is the stop-out, and since 1998 every winner pays it — the same yield, whether they bid aggressively or not. That is a single-price or uniform auction.
Uniform pricing is worth a moment. Under the older discriminatory format, each winner paid their own bid, which punished anyone who bid too aggressively — the “winner's curse.” Fear of that curse made bidders shade their bids down, which cost the Treasury money. Paying everyone the stop-out removes the penalty for bidding honestly, and honest bids are worth more to the issuer than cautious ones.
Run The Auction
A fixed ladder of competitive bids. Raise the size the Treasury needs to sell and watch the stop-out walk up the rungs; cut demand and watch the tail open up.
Run the auction
The thing to notice: the Treasury always sells all of it. There is no size on this dial at which the auction fails. What changes is the yield — and every basis point of it is paid on the whole issue, every year, until it matures.
How To Read
An Auction
Three numbers get published within minutes of the close, and between them they are how the market grades a government.
- 1The tail. Stop-out minus when-issued. Positive is weak, negative — “stopped through” — is strong. This is the headline.
- 2Bid-to-cover. Total bids divided by the amount sold. A ratio of 2.5 means two and a half times as much money wanted in as there was paper. It is a crude measure — a big number with a big tail still means weak demand, since the bids were there but only at punishing yields — but a collapsing bid-to-cover is a real signal.
- 3The allotment split. Results break demand into primary dealers, direct bidders, and indirect bidders. Indirects are largely bids submitted through dealers on behalf of others, foreign central banks prominently among them. A high dealer share is the weak outcome: dealers are obliged to bid in every auction, so paper landing on their balance sheets means nobody else wanted it at that price.
Primary dealers must submit reasonable bids at every single auction. That is the price of the designation, and it is what guarantees a Treasury auction can never actually fail for lack of bidders. The government always sells the debt. What varies is the yield it pays — which is precisely the disciplining mechanism Module 01's wall was built to preserve. An auction cannot fail, but it can hurt.
Recap &
Your Turn
Module 06 in 6 lines:
- 1Announce, trade when-issued, auction, settle. Each step solves a problem an earlier version of the process had.
- 2The when-issued market prices a bond that does not exist yet — and that price is what makes the auction gradeable.
- 3Competitive bids state a yield and discover the price; noncompetitive bids take whatever clears and are filled first, off the top.
- 4Single-price since 1998: every winner pays the stop-out, which removes the winner's curse and buys the Treasury more honest bidding.
- 5Tail, bid-to-cover and the dealer share are the three published numbers, and the dealer share is the one that quietly says the most.
- 6Primary dealers must bid at every auction, so an auction cannot fail for lack of bidders. It can only get expensive.
Answer before you peek
Question: Under the old discriminatory format each winner paid their own bid, so the Treasury captured more from the most aggressive bidders. Uniform pricing gives that surplus away. Why is it still the better deal for the government?
Test Yourself
Five questions.
The Floor System
Module 05 ended with a puzzle. The old way of setting interest rates depended on reserves being scarce, and after 2008 they stopped being scarce. So how does the Fed move a market rate at all now?
The answer is a genuinely elegant piece of engineering, and it has a visible seam in it that teaches more than the clean version would.
The Way It
Used To Work
Before 2008, setting the interest rate was a quantity problem, and it worked beautifully because reserves were scarce.
The system held perhaps twenty billion dollars of reserves in total. Banks genuinely needed them to settle payments and to meet reserve requirements, so the demand curve for reserves was steep: take a couple of billion out and banks would bid hard for what was left, and the overnight rate would rise. Put a couple of billion in and it would fall. The Desk in New York did small repo operations every morning and the fed funds rate went where the FOMC had said it should.
That is a corridor system, and it depends entirely on scarcity. Take the scarcity away and the lever snaps off in your hand.
After 2008 the balance sheet grew by trillions, and reserves went from twenty billion to several thousand billion. Adding or draining ten billion in that world does precisely nothing — the demand curve is flat out there, because no bank is short. The Fed had to keep an enormous balance sheet for other reasons and simultaneously needed to be able to raise rates. Those two requirements are incompatible under a corridor system.
Pay Them
Not To Lend
So the Fed stopped trying to control the quantity and started controlling the price. The tool arrived in October 2008: authority to pay interest on reserves.
The logic is one sentence. No bank will lend money at less than it can earn risk-free from the Federal Reserve. If the Fed pays 4.40% on reserve balances, a bank offered 4.20% in the market says no and leaves the money at the Fed instead. The rate the Fed pays becomes a floor under every other overnight rate, and the Fed can raise or lower it by announcement, with a balance sheet of any size at all.
This is the floor system, and it is why the Fed can hold trillions in assets and still put the overnight rate exactly where it wants it. Quantity and price were divorced, which is the single most consequential operational change in modern central banking.
Why The Floor
Leaks
Except the floor leaks, and the leak is instructive.
The effective fed funds rate — the thing the FOMC actually targets — persistently trades below IORB. If no bank lends below what the Fed pays, how?
Because the biggest lenders in the fed funds market are not banks. They are the Federal Home Loan Banks, which hold accounts at the Fed but are not eligible to earn interest on reserves. Their cash has to earn something, and fed funds is where they put it. They will happily lend at 4.33% because their alternative is zero.
Which sets up an arbitrage that looks like free money: a commercial bank borrows from an FHLB at 4.33%, leaves the cash at the Fed, and earns 4.40%. Seven basis points for nothing.
Because it is not free. Borrowing that cash inflates the bank's balance sheet, and a bigger balance sheet consumes leverage-ratio capacity and attracts a larger FDIC assessment. Those costs are small but real, and they are exactly what keeps the gap open. The spread between EFFR and IORB is, quite literally, the market price of bank balance-sheet space — which is why it widens at quarter-ends when banks are managing their reported size.
// Why does the fed funds rate trade BELOW the rate the Fed pays banks? // Because the biggest lender in that market is not allowed to earn it. const IORB = 4.40; // paid by the Fed, to banks only const ONRRP = 4.25; // paid by the Fed, to money funds and GSEs function willLendAt(institution, rate) { if (institution.canEarnIORB) return rate >= IORB; // else just hold reserves if (institution.canUseONRRP) return rate >= ONRRP; // else just lend to the Fed return rate >= 0; // nowhere else to go } // Federal Home Loan Banks are the punchline. They hold Fed accounts but // are NOT eligible for interest on reserves, and their cash has to earn // something, so they lend it into fed funds below IORB. const fhlb = { canEarnIORB: false, canUseONRRP: false }; // A bank borrows from them at 4.33 and earns 4.40 doing nothing at all. // Free money — except it is not quite free: the borrowed cash inflates // the bank's balance sheet, which costs leverage-ratio capacity and an // FDIC assessment. Those frictions are the whole 7bp gap. willLendAt(fhlb, 4.33); // => true. And so EFFR sits at 4.33, not 4.40.
The Whole Stack
One more hole to plug. Money market funds hold trillions and cannot earn IORB either, because they are not banks. Left alone, they would lend at whatever they could get, dragging short rates below the Fed's floor.
So the Fed built them a door: the overnight reverse repo facility. A money fund lends cash to the Fed overnight, takes Treasury collateral, and earns the ON RRP rate. Nobody with access lends below it, because the Fed will always take the money. That is a floor for the non-banks, and it sits a little below IORB.
Take-up tells you a lot about the system's state. When there is more cash sloshing around than the private market wants, ON RRP balances swell — they exceeded two trillion dollars at the peak in 2022–23 — and every dollar parked there is a dollar not held as bank reserves. Look back at Module 05's simulator: ON RRP is one of the sliders competing for the same pool.
Above the floor sit two ceilings. The discount window has always been one in theory, but borrowing there carries a stigma so severe that banks avoid it in exactly the moments it exists for. So in July 2021 the Fed added the standing repo facility: borrow cash against Treasuries at a fixed rate, no stigma, always open. A ceiling that people will actually use, built after September 2019 demonstrated why one was needed.
Build The Box
Move the two administered rates and the friction, and watch where the market rate is forced to settle — and when it escapes the target range entirely.
Set the administered rates
Target range is 4.25%–4.50% throughout.
The thing worth provoking: set IORB near the top of the range and push the friction up. EFFR falls out of the bottom of the target range, and the Fed has publicly missed its own target — which is not a hypothetical. Managing exactly this is what the Desk does every day.
Recap &
Your Turn
Module 07 in 6 lines:
- 1The pre-2008 corridor worked because reserves were scarce — small operations moved the rate a long way.
- 2Trillions of reserves flattened the demand curve and broke that lever completely.
- 3Since October 2008 the Fed pays interest on reserves, which divorces the price of money from the quantity of it.
- 4EFFR still trades below IORB because Federal Home Loan Banks lend into fed funds and cannot earn IORB themselves.
- 5The arbitrage is not free: balance-sheet costs keep the gap open, and that gap is the price of bank balance-sheet space.
- 6ON RRP is the floor for non-banks, the discount window is a stigmatised ceiling, and the standing repo facility is the ceiling built after 2019 that people will actually use.
Answer before you peek
Question: The Fed pays tens of billions of dollars a year in interest to commercial banks for holding reserves they cannot collectively get rid of. Put the strongest case that this is indefensible — then answer it.
Test Yourself
Five questions.
Remittances:
The Money Sent Home
This module connects the two institutions by their income statements, and it is where the clean separation from Module 01 gets complicated — not by anybody breaking a rule, but by arithmetic.
It is also the best single answer to “who pays for quantitative easing?”, which is a question most people assume has no answer.
An Institution That
Pays For Itself
Module 01 said the single biggest structural source of Fed independence is that nobody appropriates its budget. This module is that sentence, examined.
The Fed is one of very few federal institutions that funds itself, and it does so the way a bank does: it owns interest-bearing assets. Trillions of Treasury securities and mortgage-backed securities throw off coupons. Out of that income the Fed pays its own costs, and everything left over is sent to the Treasury — weekly, automatically, with nobody voting on it.
For most of the last century those remittances were a pleasant footnote. In the QE years they became one of the largest single non-tax revenue items in the federal budget: on the order of a hundred billion dollars a year. A central bank was, in effect, quietly one of the government's better-performing revenue sources.
Then in 2022 it went to zero, and stayed there.
Where The Money Goes
Six lines, in strict order.
The Year It
Went Negative
The mechanism is an interest-rate mismatch, and once you see it you cannot unsee it.
The Fed's assets are long-dated bonds, most of them bought between 2020 and 2021 when yields were close to zero. A bond's coupon does not change. That income was locked in for years the moment the purchase settled.
The Fed's liabilities are reserves and reverse repo — overnight money — and Module 07 explained that the Fed pays interest on them at the policy rate. That cost reprices the day the FOMC votes.
So when the FOMC raised rates rapidly through 2022, the payout side leapt while the income side sat still. Somewhere in September 2022 the two lines crossed, and the Federal Reserve began losing money — not on a valuation basis, but in actual cash out the door.
Not what a company does. The Fed does not raise capital, does not ask Congress for an appropriation, and cannot become insolvent in a currency it issues. It books a deferred asset: an accounting entry representing earnings it expects to make in future, against which the current shortfall is charged. Remittances stop until the deferred asset is worked off in full. It has since run into the hundreds of billions.
Whether that entry is a real asset is a genuinely contested accounting question. What is not contested is the fiscal consequence: a revenue line worth roughly a hundred billion a year disappeared from the federal budget, and will not return until the hole is filled.
function annualRemittance(fed) { const earned = fed.portfolio * fed.portfolioYield; // locked in years ago const paidOut = (fed.reserves + fed.onRRP) * fed.policyRate; // reprices overnight const net = earned - paidOut - fed.operatingCosts - fed.dividends; if (net > 0) return { toTreasury: net, deferredAsset: 0 }; // A central bank cannot be insolvent in its own currency, and it does not // go to Congress for capital. It books the shortfall as a "deferred asset" // — an IOU to itself against future profits — and remits nothing until // the whole thing is worked off. return { toTreasury: 0, deferredAsset: -net }; } // The asymmetry is the whole module. The left-hand term is a portfolio of // long-dated bonds bought when yields were near zero: it cannot reprice. // The right-hand term is overnight money: it reprices the day the FOMC votes. annualRemittance({ portfolio: 7000, portfolioYield: 0.023, // bought in 2020-21 reserves: 3200, onRRP: 400, policyRate: 0.048, operatingCosts: 9, dividends: 1 }); // => { toTreasury: 0, deferredAsset: ~17 } ...per year, and it compounds.
Run The Income Statement
Set the portfolio the Fed is stuck with, and the rate it has to pay tonight.
Run the Fed's income statement
Try the historically honest configuration: a portfolio yielding around 2% because it was bought in 2020, against a policy rate near 5%. Then drag the policy rate back down and watch how long it takes the deferred asset to start shrinking. That lag is the fiscal shadow of a monetary decision, and nobody voted for it as a budget item.
Consolidate The Two
And Look Again
Now step back and consolidate, because this is where the module earns its place in the course.
The Treasury pays interest on its debt. A large share of that debt is owned by the Federal Reserve. The Fed's profits go back to the Treasury. So on the slice the Fed owns, the government is — in normal times — paying interest to itself and getting it back.
Which invites the consolidated question: what has actually happened to the government's obligations? The Fed bought a long-dated fixed-rate Treasury bond and paid for it by creating reserves, on which it pays the overnight policy rate. Look at the government as a single entity and that transaction did something specific: it retired a long-term fixed-rate liability and replaced it with an overnight floating-rate one.
That is a maturity swap on the national debt, executed by an institution with no fiscal mandate, as a side effect of monetary policy. In a low-rate world it is nearly free and it generates enormous remittances. In a high-rate world it costs real money, immediately, and the remittances vanish. Nothing went wrong; the trade simply had two sides and the second one arrived.
This is the strongest version of the argument that quantitative easing has fiscal consequences, and notice how little it needs. No claim that the Fed “printed money to fund the deficit,” no conspiracy, no violation of the wall from Module 01 — every bond was bought second-hand at a market price. The fiscal effect falls straight out of double-entry bookkeeping on a consolidated balance sheet, and the people who ran the policy were explicit about it at the time.
Recap &
Your Turn
Module 08 in 6 lines:
- 1The Fed takes no appropriation. It earns interest on its portfolio, pays its costs, and remits the residual to the Treasury weekly.
- 2In the QE years that residual ran near $100bn a year — one of the larger non-tax revenue lines in the federal budget.
- 3Its assets are long-dated and fixed; its liabilities are overnight and floating. The two sides reprice at completely different speeds.
- 4When the FOMC raised rates fast in 2022, the payout side crossed the income side and remittances stopped.
- 5A central bank that loses money books a deferred asset rather than raising capital, and remits nothing until it is worked off.
- 6Consolidated, QE retired long-term fixed-rate government debt and replaced it with overnight floating-rate debt. That is the fiscal side of a monetary policy.
Answer before you peek
Question: The Fed's losses do not threaten its solvency and it can always pay its bills. So why does anyone care about the deferred asset at all?
Test Yourself
Five questions.
QE, QT & Who
Funds A Deficit
This is the most argued-about topic in the course and the one where the popular version and the technical version have least in common.
Modules 05 and 08 did the hard preparation: you know what a reserve is, you know the Fed's assets are long and its liabilities are overnight, and you know remittances are the residual. With those three facts the whole debate becomes legible — including which side of it is right about what.
Two Wrong Answers
Quantitative easing gets described two ways, and both are wrong.
The first is “the Fed printed money and gave it to the government.” Module 01 already ruled that out: the Fed cannot buy from the Treasury, so every bond it owns was bought from an investor who already held it, at a price the market set. No dollar of federal spending was financed by a Fed purchase in any direct sense.
The second is “QE flooded the economy with money.” Also not quite. What QE creates is reserves, and Module 05 was emphatic that reserves are not deposits, cannot leave the banking system, and are not spendable by anyone outside it. The pension fund that sold the bond does get a deposit — but it also gave up a bond of identical value. Nobody got richer.
So what does it actually do?
One Purchase,
Step By Step
Three Channels,
Three Confidence Levels
Three things, in descending order of how confident anyone should be about them.
- 1It removes duration from the market. This is the mechanism people are most confident about. Long-dated bonds carry interest-rate risk; somebody has to hold it. When the central bank takes trillions of that risk onto its own balance sheet, the private sector is left holding less, and the compensation demanded for holding it — the term premium — falls. Long rates come down even though the overnight rate has not moved.
- 2It signals. Buying long bonds is an expensive, visible commitment that short rates will stay low for a long time — because the central bank has just given itself a large position that would lose value if they did not. The signal works precisely because it is costly.
- 3It repairs broken markets. The least glamorous channel and possibly the most important. In March 2020 the Treasury market itself stopped functioning — the deepest market in the world could not absorb sellers. The Fed bought enormous quantities in a few weeks for no monetary reason at all: it was acting as buyer of last resort to keep the plumbing open. This is a different policy that happens to use the same button.
“QE” names at least three policies with different goals, different evidence bases and different exit paths. Market-functioning purchases should be unwound quickly and largely were. Duration-extraction purchases are a long-horizon bet. Signalling is nearly free until the signal has to be broken. Most arguments about whether QE “worked” are two people evaluating different channels.
The Same Debt,
A Different Shape
Module 08 introduced this move. Here it is drawn.
Do The QE Yourself
Buy long bonds with created reserves and watch what happens to the maturity of the government's consolidated debt — and to its annual interest bill as you move the overnight rate.
Consolidate the two balance sheets
Fixed: $28,000bn of debt held by the public, average maturity 6 years at a 3.0% average coupon before any purchases.
Set the purchases high and the overnight rate near zero: the consolidated interest bill falls, and remittances are enormous. Now drag the rate to 5% without touching anything else. Nothing about the policy changed and nothing was done wrong — the position simply has two sides.
Unwinding, Slowly
Unwinding is deliberately duller than easing, and the dullness is the design.
The Fed does not sell. It lets securities mature and run off, replacing less than it holds, subject to monthly caps: up to a set amount rolls off, anything above the cap is reinvested. Slow, pre-announced, and mechanical — Module 02's regular-and-predictable doctrine, borrowed by the other institution.
The reason for passivity is Module 05's arithmetic. Every dollar of runoff destroys a dollar of reserves, and nobody knows exactly where “ample” ends. Selling actively would mean discovering that boundary at a speed of the Fed's choosing rather than the market's — and September 2019 is what finding it unexpectedly looks like. Running off slowly means you can stop when the funding markets start to twitch.
The Argument
That Survives
Which brings us to the argument that actually has teeth, and it is not the one people usually make.
Set aside money-printing. Consolidate the government — Treasury and Fed as one entity, as Module 08 did — and QE looks like this: the state retired long-dated fixed-rate obligations and issued overnight floating-rate ones in their place. The debt did not grow. It got shorter.
Everything follows from that. A government whose liabilities float overnight feels a rate rise immediately and completely, instead of over a decade. It earns a great deal when rates are low — those hundred-billion-dollar remittances — and pays a great deal when they are high. And the institution that performed this maturity transformation on the national debt did so as a side effect of pursuing an inflation target, with no fiscal mandate and no vote in Congress on the debt-management implications.
The name for the far end of this worry is fiscal dominance: a situation where the fiscal consequences of monetary decisions become large enough that the central bank cannot make them freely any more. If raising rates by two points blows a hole in the budget and eliminates a major revenue line, a central bank is no longer choosing between inflation and employment alone. Nobody serious argues the United States is there. Plenty of serious people argue the distance has shrunk, and this is the mechanism by which it shrank.
Recap &
Your Turn
Module 09 in 6 lines:
- 1QE does not fund the government directly — the open-market wall holds, and every bond was bought second-hand at a market price.
- 2It does not put spendable money in the economy either: it creates reserves, and the seller swapped a bond for a deposit of equal value.
- 3What it does is take duration out of private hands, signal that short rates will stay low, and — in March 2020 — keep a broken market open.
- 4QT is passive runoff under monthly caps, because nobody can locate the lower edge of 'ample' and September 2019 showed what finding it costs.
- 5Consolidated, QE shortened the maturity of government debt: long fixed-rate obligations retired, overnight floating-rate ones issued.
- 6Fiscal dominance is the far end of that: when the budget consequences of a rate decision get big enough to constrain the decision itself.
Answer before you peek
Question: Suppose the Fed had bought exactly the same quantity of bonds, but the Treasury had simultaneously lengthened its issuance enough to offset the maturity effect. Would the consolidated criticism in 09.7 still apply — and what does your answer imply about who is actually responsible for the debt's maturity?
Test Yourself
Five questions.
The Debt Ceiling
& The TGA
Everything in this module is a consequence of one fact established in Module 01: the government's checking account is a liability of the central bank.
If the TGA were held at a commercial bank, a debt-ceiling standoff would be a purely fiscal and legal drama. Because it is held at the Fed, it is also a monetary event of considerable size — running in the opposite direction to what almost everyone assumes.
Not A Limit
On Spending
The debt ceiling is the most misunderstood instrument in American public finance, and almost all of the confusion comes from one point.
It is not a limit on spending. Congress has already voted for the spending, separately, in appropriations bills. The ceiling is a limit on borrowing to pay for things Congress has already ordered. Hitting it does not cancel any obligation; it removes the means of settling them.
The closest household analogy is not deciding to spend less. It is having already bought the things, and then cancelling your own overdraft.
Historically, as a convenience. Before 1917 Congress authorised each bond issue individually, which was impractical for a country fighting a world war. The aggregate limit was created to give the Treasury flexibility — a delegation of authority, not a restraint on it. It became a point of leverage much later, and that is a use it was never designed for. Denmark has something similar and raises it without drama; almost no other country has the instrument at all.
Extraordinary Measures
When the ceiling binds, the Treasury does not stop paying. It starts a long, well-rehearsed sequence of accounting manoeuvres known as extraordinary measures.
Broadly, these involve temporarily disinvesting or under-investing certain government trust funds — federal employee retirement accounts prominently — which frees up borrowing capacity under the cap. The funds are made whole with interest afterwards, and it is entirely legal and entirely routine. What it buys is time: typically a few months.
Then the Treasury also does the thing that connects this module to the whole course. It spends down the cash in the TGA, because that is the one source of money that does not require issuing debt.
And now look back at Module 01. The TGA and bank reserves are two rows in the same column of the Fed's balance sheet. Draining the government's account pushes reserves into the banking system, hundreds of billions of them, precisely while the political news is at its most alarming. A fiscal crisis produces a monetary easing that nobody voted for and nobody wanted.
The X-Date
The X-date is the day the measures and the cash both run out — the first day the government cannot pay everything it owes.
Estimating it is a cash-flow forecasting problem, and it is genuinely hard for a reason worth understanding: federal receipts are not smooth. They arrive in enormous lumps on a handful of days. Move a single quarterly corporate tax date across the boundary and the X-date can shift by weeks. This is why Treasury announces a range rather than a date, why the range keeps moving, and why “we will know more after April receipts” is a substantive answer.
// The single number every analyst on Wall Street tries to estimate during // a standoff. It is a cash-flow forecast, and it is genuinely hard. function daysUntilX(cash, headroom, dailyNet) { let day = 0; while (cash + headroom > 0) { cash += dailyNet[day % dailyNet.length]; // wildly uneven, by design headroom = Math.max(0, headroom - 0); // burns down as measures run out day++; if (day > 400) return "beyond the forecast horizon"; } return day; } // Why it is hard: dailyNet is not a smooth series. It is a calendar of // enormous lumps — a benefit payment day here, a corporate tax date there, // a coupon settlement on the 15th. Move one large receipt by a week and // the X-date moves by a fortnight. // // Which is why Treasury announces a RANGE, why the range keeps changing, // and why "we will know more after April receipts" is a real answer and // not evasion.
Find The X-Date
A stylised cash-flow model. Note how violently the answer moves when the tax date crosses the boundary — that sensitivity is why the real estimate is a range.
Find the X-date
Both Sides
Of The Episode
The two lines you now know move in exact opposition.
The resolution is where the monetary consequence actually bites, and hardly anyone is watching by then.
The moment the ceiling is raised, the Treasury has to rebuild a cash balance it ran to almost nothing — back to the buffer policy from Module 02, which means several hundred billion dollars. It does that by issuing a wall of bills in a matter of weeks.
Every dollar of that lands in the TGA and comes out of bank reserves. Module 01's dial, spun hard, in one direction, fast. It is a monetary tightening of a size the FOMC would deliberate for months over, delivered as an administrative consequence of a political settlement. In 2023 this was large enough that the Fed's own communications discussed it, and money-market participants spent the summer watching a fiscal calendar rather than a monetary one.
The Workarounds
Two workarounds get proposed every cycle, and both are more interesting than they first appear.
The platinum coin
A statute authorising platinum commemorative coins in any denomination the Treasury chooses. Mint a coin, deposit it at the Fed, credit the TGA — no borrowing, no ceiling.
It has never been used. The objection is not really legal; it is that a government which settles a constitutional dispute by exploiting a coin-collecting statute has told the world something about how seriously to take the rest of its commitments. The Fed has also signalled it would not want to be the counterparty.
Premium bonds
The ceiling limits the face value of debt. Issue a bond with a far above-market coupon and it sells for well over par — you raise more cash per dollar of face value counted against the cap.
Cleaner, technically legal, and clearly against the spirit of the limit. It also costs more in interest over the life of the bond, so it buys headroom with real money.
Which leaves prioritisation — paying bondholders first and delaying everything else. The Treasury has consistently said its systems are built to pay bills in the order they come due and that selective payment is not straightforwardly operable. Whether it is possible at all is disputed; that nobody wants to find out is not.
Recap &
Your Turn
Module 10 in 6 lines:
- 1The ceiling limits borrowing for spending Congress has already ordered. It cancels no obligation — it removes the means of settling them.
- 2It began in 1917 as a convenience, delegating flexibility to the Treasury. Its use as leverage came much later and was never the design.
- 3Extraordinary measures buy months by temporarily under-investing certain trust funds. Entirely legal, entirely routine.
- 4The Treasury also drains the TGA — which pushes reserves INTO the banking system. A fiscal crisis produces a monetary easing.
- 5The X-date is hard to forecast because receipts arrive in lumps; one tax date crossing the line moves it by weeks.
- 6The tightening comes after the deal, when the Treasury rebuilds its buffer with a wall of bills and drains reserves back out.
Answer before you peek
Question: Almost everyone assumes a debt-ceiling standoff tightens financial conditions. On the reserve plumbing it does the opposite. Reconcile those two things — and say which one an investor should actually care about.
Test Yourself
Five questions.
Emergency Powers:
13(3) & The ESF
Every module so far has described a machine operating normally. This one is about the override switch — and about a design question that is more interesting than the powers themselves.
The question: an institution that can create unlimited money and is deliberately insulated from elections is exactly the wrong body to decide who gets rescued. So how do you let it act in a crisis without letting it make that decision?
Unusual And Exigent
Everything so far has been the Fed dealing with banks. Section 13(3) of the Federal Reserve Act is what happens when the problem is not at a bank.
In “unusual and exigent circumstances,” and with the agreement of at least five of the seven governors, the Board may authorise lending to institutions that are not banks at all — securities dealers, corporations, municipalities, whoever the emergency happens to involve. It is the broadest power the Fed has, it sat unused for roughly seventy-five years after the Depression, and then it was used constantly for two years.
2008 — targeted and broad
Facilities for commercial paper, asset-backed securities, primary dealers — plus rescues aimed at individual firms, which is the part that generated the lasting political damage. Bear Stearns and AIG were not programmes; they were interventions in named companies.
2010 — Congress narrows it
Dodd-Frank rewrote 13(3): lending must now be through a broad-based programme rather than aimed at rescuing one firm, and — critically — the Treasury Secretary must approve it. The elected side of the government was written into the emergency power.
2020 — the new shape
Facilities for corporate credit, municipal debt, mid-sized business lending and the payroll programme. All broad-based, all approved by the Secretary, and all layered on Treasury capital appropriated by Congress in the CARES Act.
Who Takes
The Credit Risk
That last point is the one worth slowing down for, because it is the most elegant piece of institutional design in the whole course.
The Fed can create unlimited liquidity. What it cannot do — should not do — is decide which private firms are worth exposing the public to losses on. That is a distributional judgement, and distributional judgements belong to people who stand for election.
So the modern emergency facility is built in layers. Congress appropriates money. The Treasury puts that money into the facility as first-loss equity. The Fed then lends a multiple of it. If the assets sour, the losses eat the Treasury's equity first, and only reach the Fed's loan if the equity is entirely gone.
Read that structure as a division of labour and it is exact. Liquidity is the Fed's job and it has an infinite supply. Credit risk is a fiscal choice and it has to be appropriated. The Fed stays a lender rather than an investor; Congress stays the body that decides who gets public money; and the accountability for any losses lands somewhere a voter can reach.
The Exchange
Stabilization Fund
There is one more pot of money in this story, and it is genuinely strange.
The Exchange Stabilization Fund was created by the Gold Reserve Act of 1934 out of the profits the government booked when it revalued gold. It belongs to the Treasury, it is controlled by the Secretary, and — this is the unusual part — it operates largely outside the annual appropriations process.
Its original purpose was intervening in currency markets. Over time it became the closest thing the executive branch has to discretionary financial firepower: it backstopped Mexico in 1995 when Congress would not, guaranteed money market funds in 2008 when they were breaking, and supplied the initial equity for several 2020 facilities before the CARES appropriation arrived.
It is worth knowing about for the same reason the deferred asset was: it is a channel through which fiscal capacity moves without a vote. Small compared to the federal budget, decisive at the moment it is used, and almost invisible until then.
The Loss Waterfall
Three layers, and the order is the whole point.
Blow Up A Facility
Size the facility, size the equity underneath it, then decide how badly the assets perform.
Blow up a facility
Two things to try. Set the loss rate to zero — which is roughly what happened to the 2020 facilities — and notice that the Fed made money and the Treasury equity came back untouched. Then push the loss rate past the equity and watch the last line change: the central bank starts absorbing credit losses on assets it chose, which is the situation the whole structure exists to prevent.
Two Limits
Worth Stating
Two limits are worth stating plainly, because emergency powers are the place where “the Fed can do anything” gets said most often.
First, 13(3) is a lending power, not a spending power. The Fed makes loans against collateral and expects repayment. It cannot make grants, and the statute requires security it judges satisfactory. When the 2020 programmes needed to accept real credit risk, that risk had to be appropriated by Congress and carried by the Treasury — which is exactly the layering above.
Second, the most powerful thing about these facilities has repeatedly turned out to be their existence rather than their use. Several 2020 programmes were announced, moved markets immediately, and then lent remarkably little. A credible commitment that a buyer will be there changes behaviour without the buyer having to show up. That is a real and slightly uncomfortable form of power, because it is enormous and it leaves almost no footprint on any balance sheet.
Recap &
Your Turn
Module 11 in 6 lines:
- 1Section 13(3) lets the Fed lend to non-banks in unusual and exigent circumstances, with five of seven governors agreeing.
- 2It sat unused for about seventy-five years, then carried much of the 2008 and 2020 responses.
- 3Dodd-Frank narrowed it in 2010: broad-based programmes only, no more rescues of named firms, and the Treasury Secretary must approve.
- 4Modern facilities layer Treasury first-loss equity under a Fed loan — liquidity is the Fed's job, credit risk is a fiscal choice that must be appropriated.
- 5The Exchange Stabilization Fund is executive-controlled financial firepower outside the annual appropriations process, decisive whenever it is used.
- 613(3) is a lending power, not a spending power — and the announcement of a facility has repeatedly done more work than the lending did.
Answer before you peek
Question: The 2020 corporate credit facilities were announced, moved markets enormously, and then bought comparatively little. Is that a success or a problem?
Test Yourself
Five questions.
Independence:
1951 To Now
Eleven modules of mechanism. This one is the argument.
Everything you now know about the plumbing exists inside a political settlement that was won in a specific fight, in 1951, and has been contested at intervals ever since. The settlement is not in the Constitution, it is barely in the statute, and it is considerably more fragile than the machinery it protects.
When There Was
No Independence
Start where the independence we have been describing did not exist, because it makes everything afterwards legible.
During the Second World War the Federal Reserve agreed to peg the yield on Treasury debt — around three-eighths of a percent on bills, two and a half on long bonds — so that the government could finance the war cheaply. To hold a yield you must buy whatever quantity the market will not absorb at that price. Which means the Treasury decided how much to issue, and the Fed's balance sheet became the residual.
That is not a central bank with independence in any meaningful sense. It is a financing operation with a research department. And through the war it was defensible — there was an obvious national emergency and an agreed objective.
The trouble came afterwards. The war ended, the peg stayed, and by 1950 inflation was running hard. The Fed wanted to raise rates; the Treasury insisted the peg continue. The dispute went to the President. What broke it was, in the end, an institutional refusal — and the resulting Treasury–Fed Accord of March 1951, a short and rather unglamorous document releasing the Fed from the obligation to support the government's borrowing costs.
Everything in this course after Module 01 depends on that document. It is the moment the separation stopped being formal and became real.
Six Times It
Was Tested
Six times it was tested
Click an episode. Each one is a different answer to the same question: what happens when the government wants cheaper money than the central bank thinks is wise?
Where It Stands
The pressure never really stops; what varies is how visible it is and how the institution responds. A few things are worth saying about the present moment carefully.
The legal protection around a governor's tenure is a phrase — removal “for cause” — that Congress never defined and that went untested for most of a century. It has recently been tested, and the durability of that protection is now an open question working through the courts rather than a settled assumption. I am not going to tell you how it comes out; anyone who does is guessing.
What is worth noticing is which protection is under pressure. Module 01 argued that the load-bearing one is not tenure but funding — an institution nobody appropriates for cannot be squeezed annually. Tenure protects individuals; self-funding protects the institution. A challenge to removal is a challenge to particular people, and the Board has fourteen-year staggered terms precisely so that outlasting an administration is the normal case.
The more consequential pressure may be the one from Module 08. An institution that is visibly making losses, remitting nothing to the Treasury, and paying tens of billions to commercial banks is in a much weaker rhetorical position than one quietly handing over a hundred billion a year. None of that changes the economics. All of it changes the politics.
What The Evidence
Actually Shows
So is independence actually good? The honest answer is that the evidence is strong but narrower than its advocates usually claim.
The core empirical finding, replicated across many countries and decades, is that more independent central banks are associated with lower and more stable inflation, without a systematic cost in growth or unemployment. That is about as close to a free lunch as macroeconomics offers, and it is why nearly every country adopted the model between the 1980s and the 2000s.
The theoretical reason is a commitment problem. A government that can set interest rates always has a short-run reason to set them too low — elections arrive sooner than inflation does. Everyone knows this, so everyone expects it, so expected inflation rises, so the government has to run tighter policy than it would have needed to if it could simply have promised not to cheat. Handing the decision to someone who cannot be replaced before the next election is a way of making the promise credible. The independence is not there because central bankers are wiser. It is there because the promise is worth more than the discretion.
The Case Against,
At Full Strength
The case against deserves its strongest form, not a caricature, because it has grown considerably better since 2008.
- 1The mandate has expanded far beyond what the argument covers. The commitment story justifies insulating a technical decision about an interest rate. It says much less about buying five trillion dollars of assets, choosing which credit markets get a backstop, or deciding whether to hold mortgage bonds rather than Treasuries. Those are allocative choices with visible winners, and the case for insulating them from politics is weaker.
- 2Monetary policy has distributional consequences. Low rates lift asset prices, and assets are held disproportionately by people who already have them. High rates raise unemployment, which falls disproportionately on people who do not. Calling these “side effects” of a technical decision is a choice about framing, not a fact.
- 3Accountability is thin where it matters most. Transcripts after five years and semiannual testimony are real, but they operate long after the decision. Five of the twelve votes were never confirmed by anyone, and Module 11 showed that the most powerful interventions can leave almost no record at all.
- 4The fiscal entanglement from Modules 08 and 09 is new. When monetary decisions swing the federal budget by a hundred billion a year and quietly reshape the maturity of the national debt, the clean line between monetary and fiscal has stopped describing reality — whatever the statute says.
The reasonable position, and the one most economists who study this actually hold, is that the empirical case for instrument independence on the interest rate remains strong, and that the governance arrangements for everything else the modern Fed does have not caught up with what it now does.
Where This Leaves Us
Twelve modules ago this started with a puzzle: the government can create dollars, and the government can run out of money, and both are true.
You now know exactly why. One institution spends and borrows and cannot create; the other creates and cannot spend. They meet at a single checking account on a central bank's liability side, and that adjacency is enough to make every fiscal action a monetary event and every monetary action a fiscal one. The wall between them is one clause about open-market purchases, and it is thinner than its importance would suggest.
What holds the arrangement up is not really any of the legal machinery. It is a norm, roughly seventy-five years old, that the government does not tell the central bank what the interest rate should be, and the central bank does not decide who gets public money. Norms are not self-enforcing. They persist because enough people in enough institutions keep choosing to act as though they bind — which is a less reassuring foundation than a statute, and an accurate description of most of what actually governs a state.
That is the course. Go and read a Daily Treasury Statement; you will find you can.
Recap &
Your Turn
Module 12 in 6 lines:
- 1Through the 1940s the Fed pegged Treasury yields — the Treasury decided issuance and the Fed's balance sheet was the residual. That is not independence.
- 2The Treasury–Fed Accord of March 1951 ended the peg and is the moment the separation became real. Everything in this course rests on it.
- 3Pressure has recurred in every era; what varies is visibility. Tenure protects individuals, self-funding protects the institution.
- 4Cross-country evidence links independence to lower, more stable inflation with no systematic growth cost — because the commitment is worth more than the discretion.
- 5The case against has strengthened since 2008: an expanded mandate, visible distributional effects, thin real-time accountability, and genuine fiscal entanglement.
- 6What holds it up is a norm about seventy-five years old, not a constitutional guarantee — which is a less reassuring foundation and an accurate one.
Answer before you peek
Question: You have the whole machine now. Final question, and it is the one the course was built to let you answer properly: if you were designing this from scratch today, would you keep the Fed's self-funding?
Test Yourself
Five to close it out.
Twelve Modules,
One Dollar
You started with a sentence that sounded like a contradiction: the United States government can create dollars, and the United States government can run out of money. You can now say exactly which institution each half of that refers to, why they were separated, where the seam is, and what happens to the plumbing when the politics goes wrong.
The best test of whether it landed is not the quizzes. It is that a Daily Treasury Statement, an FOMC statement, and an auction result are now three documents you can read for what they actually say.