Daniel sent us this one — he used to read the Financial Times religiously back when he was figuring out his career, and one of the metrics that kept catching his eye was LIBOR, the London Interbank Offered Rate. It's been phased out now, but the question that nagged at him is the one that probably hits anyone who first hears about interbank lending. Banks hold all our money, right? They're the places with the giant vaults in the popular imagination. So why on earth do they need to borrow from each other? And now that LIBOR is gone, what replaced it, and what should someone actually watch if they want to follow the interest rate chain?
It's such a good question because it exposes the gap between the cartoon version of banking and what actually happens every night while we're all asleep. The cartoon is Scrooge McDuck swimming in a vault of coins. The reality is that your deposit isn't sitting in a vault. It's a liability on the bank's balance sheet. The bank owes you that money. And at the same time, the bank has to settle payments with every other bank, every single day, in real time, with actual central bank reserves. So a bank can be perfectly solvent, have plenty of assets, and still end Tuesday afternoon short of the reserves it needs to settle what it owes to, say, JPMorgan Chase by close of business.
The bank isn't broke. It's just... momentarily in the wrong currency, effectively.
It's a timing problem. And that's the entire reason the overnight interbank market exists. Bank A has a surplus of reserves at the Fed at the end of the day. Bank B has a deficit. Bank A lends to Bank B overnight, Bank B pays it back the next morning with a tiny bit of interest, and the whole payment system doesn't seize up. That interest rate is what LIBOR was supposed to measure. It was the wholesale price of unsecured bank borrowing. Not what banks charge you for a mortgage, but what they charge each other for a one-month or three-month loan with no collateral. Just pure trust.
Which is already a fascinating phrase. The wholesale price of trust. And that trust turned out to be... let's say, a little more negotiable than anyone advertised.
That's putting it gently. So let's start with the mechanics, because to understand why LIBOR failed you have to understand how it was built. dollar LIBOR, every morning a panel of sixteen banks — the big ones, Barclays, Deutsche Bank, UBS, Citibank — would answer a question. And here's the actual question the British Bankers' Association sent them. Quote: "At what rate could you borrow funds, were you to do so by asking for and then accepting interbank offers in a reasonable market size just prior to eleven a.
"Could you borrow." Not "did you borrow.
Not a single actual trade needed to back it up. The submissions came in, the top four and bottom four were tossed out to discourage outliers, and the middle eight were averaged. That average was LIBOR. It set the price for an estimated two hundred trillion dollars in financial contracts globally. Mortgages, student loans, corporate debt, interest rate swaps — all of it keyed off a number that was, at bottom, sixteen people's educated guesses about what their bank might theoretically pay if it hypothetically went borrowing.
The most important number in global finance was essentially a very formalized survey. A very British survey. "Were you to do so." It's almost polite.
That politeness masked an enormous structural problem. If you're a bank treasurer during a crisis, and you submit a high LIBOR number, what are you telling the world? You're telling the world that other banks don't trust you enough to lend to you cheaply. You're broadcasting weakness. So during the two thousand eight crisis, LIBOR submissions were lower than the actual rates banks were paying — because no one wanted to be the bank that admitted it couldn't borrow. The rate that was supposed to signal stress in the banking system was being artificially suppressed by the very stress it was meant to measure.
That's the kind of circular failure that sounds made up when you describe it. The fear thermometer was afraid of showing a fever.
It gets worse. Because it wasn't just fear. It was profit. Starting around two thousand five, traders at multiple banks figured out that if you could nudge LIBOR up or down by even a fraction of a basis point, you could make enormous money on derivatives positions. A trader at Barclays would send a message to the person who submitted Barclays' LIBOR quote and say, effectively, "I need three-month LIBOR a little lower today, cheers." And the submitter would comply. 's Financial Services Authority found more than two hundred documented requests from Barclays traders to their own LIBOR submitters between two thousand five and two thousand nine.
This was across multiple banks, right? Not just one bad apple.
Oh, it was systemic. Barclays paid four hundred fifty-three million dollars in fines to U.regulators in twenty twelve. UBS paid one point five billion. Deutsche Bank, Royal Bank of Scotland, Citigroup, JPMorgan — the total fines globally exceeded nine billion dollars. And these weren't fines for a rogue employee. These were institutional failures where the line between the trading desk and the treasury desk had completely dissolved.
What gets me is the banality of it. These weren't elaborate heists. It was someone on a trading floor sending a message like "pls lower 3m libor today, thx." And it worked.
Because the whole edifice was built on expert judgment. And expert judgment, it turns out, is vulnerable to exactly two things: fear and greed. The Wheatley Review in twenty fourteen — that's the official U.government report on reforming LIBOR — basically concluded that a rate this important cannot rest on hypotheticals. It has to be anchored in actual transactions. But by then the damage was done. The trust was gone. And regulators decided to phase LIBOR out entirely.
The question Daniel's really asking is: what do we look at now? If LIBOR was the number, and it was a fiction, what's the reality?
replacement is called SOFR — the Secured Overnight Financing Rate. And it's about as philosophically opposite to LIBOR as you can get. SOFR is published by the Federal Reserve Bank of New York every morning at eight a.It's based on actual transactions in the Treasury repurchase agreement market — the repo market. And the volume is staggering. Over one trillion dollars in daily transactions feed into that single number. Not sixteen opinions. A trillion dollars of real trades.
Explain the repo market for someone who's heard the term but never had it laid out. What's actually happening in those transactions?
A repo is basically a very short-term collateralized loan. Imagine a hedge fund that owns a bunch of Treasury bonds and needs cash overnight. It sells those bonds to a money market fund with a promise to buy them back the next morning at a slightly higher price. The difference in price is the interest rate. The Treasury bonds are the collateral. If the hedge fund fails to buy them back, the money market fund keeps the bonds. So the loan is secured — there's an asset backing it.
That's the key difference from LIBOR. LIBOR was unsecured. If the bank went under overnight, you were just another creditor in line.
SOFR is secured by the safest collateral in the world — U.That makes it much less sensitive to bank credit risk, which is both a feature and a bug. The feature is that nobody can manipulate it by faking their own creditworthiness. The bug is that during a genuine banking panic, SOFR might not spike the way LIBOR would have, because lenders in the repo market are protected by collateral. They're not worried about the borrower's health, they're holding Treasuries. So the rate stays calm even as the banking system is quietly catching fire.
Which means we might lose the early warning signal. The canary in the coal mine just got a little soundproof booth.
That's the debate. Some economists argue we still need a credit-sensitive benchmark — something that captures what it actually costs an unsecured bank to borrow. The alternative is that we only find out about bank stress when it's already advanced enough to show up somewhere else. And "somewhere else" is often not where you're looking.
Okay, but before we get to the philosophical problem with SOFR, walk me through the actual transition. Because two hundred trillion dollars in contracts referenced LIBOR. You can't just flip a switch.
You absolutely cannot. Financial Conduct Authority stopped compelling banks to submit LIBOR quotes after December thirty-first, twenty twenty-one. And all remaining LIBOR settings — with a few synthetic exceptions — ceased after June thirtieth, twenty twenty-three. So LIBOR is dead. But here's the thing: trillions of dollars in legacy contracts still referenced LIBOR and had no built-in fallback language. If your ten-year corporate loan from twenty eighteen said "interest is three-month LIBOR plus two percent," and suddenly three-month LIBOR doesn't exist, what happens? Potentially, the contract breaks.
"the contract breaks" in the world of international finance is the kind of phrase that makes central bankers wake up at three in the morning.
It's existential. So regulators in multiple jurisdictions passed legislation to automatically transition legacy contracts to a SOFR-based equivalent. In the U., the Adjustable Interest Rate Act of twenty twenty-one did exactly that. But the conversion isn't seamless. LIBOR had term rates — one month, three months, six months, twelve months. You could lock in a rate for a specific period. SOFR is overnight. Every day, a new rate. If you have a five-year corporate loan that used to reference three-month LIBOR, you now need to construct a three-month rate out of daily SOFR readings. That's called "SOFR compounded in arrears." And it introduces what's known as basis risk — the risk that the compounded SOFR rate doesn't perfectly match the funding cost the bank actually faces.
The new number is more honest, but also more complicated to actually use in contracts that weren't designed for it.
And there's another layer. The CME Group now publishes something called Term SOFR — forward-looking SOFR rates for one month, three months, six months. But here's the catch: Term SOFR is derived from SOFR futures markets, not from actual repo transactions. So we've come full circle. We replaced a hypothetical survey with transaction data, and now we're building term rates out of derivatives markets. They're more robust, certainly — the futures market is deep and liquid — but they're still one step removed from the underlying transactions.
The purity of "only real trades" lasted about five minutes before the market needed term structure and we started building models on top of the data.
Finance always rebuilds the complexity. The question is whether the new complexity sits on a foundation of something real. And with SOFR, it does. A trillion dollars a day in real repo trades is a foundation. Whatever you build on top of it, the ground floor is solid. That's genuinely different from sixteen people answering a hypothetical question.
Let's zoom out, because Daniel's question is really about the interest rate chain. How does this wholesale rate actually flow through to what a normal person sees on their mortgage or their savings account?
The chain has several links. Step one: the Federal Reserve sets the federal funds rate target — that's the rate banks charge each other for overnight loans of reserves. Step two: the effective federal funds rate and SOFR are the actual, observed overnight rates in different markets. Step three: those overnight rates influence what banks pay for funding — if it costs more to borrow overnight, it costs more to fund the bank's lending operations. Step four: banks pass those costs through to consumers in the form of higher loan rates and, eventually, higher deposit rates.
The pass-through isn't one-to-one and it isn't instant.
Not at all. A thirty-year fixed mortgage is priced off the ten-year Treasury yield, not directly off SOFR. But SOFR influences the spread the bank charges. If a bank's funding cost rises by fifty basis points, that spread has to widen or the bank's margin gets compressed. So a sustained rise in SOFR eventually pushes up mortgage rates, auto loans, credit card rates — but with a lag and a spread that varies by product and by bank.
The September twenty nineteen repo spike is the perfect case study in how this can go wrong fast.
That was a genuine plumbing failure. September seventeenth, twenty nineteen. Overnight repo rates suddenly spiked to ten percent. Normal was around two percent. The reason was a coincidence of two things: corporate tax payments were due, which drained reserves from the banking system, and a Treasury auction settled the same day, which meant dealers had to finance large purchases of new bonds. Suddenly there was a massive demand for overnight cash and not enough supply. The repo market seized up. The effective federal funds rate briefly spiked above the Fed's target range. The New York Fed had to step in and inject liquidity — essentially, emergency overnight lending — to bring rates back down.
Even with a transaction-based rate, even with a trillion dollars in daily volume, the plumbing can still back up. The rate doesn't lie, but it can scream.
That's the trade-off. LIBOR might have lied during a crisis. SOFR tells the truth — sometimes a very alarming truth — but it's a truth about the repo market, not about bank credit risk. If the next crisis originates in the banking system rather than the repo market, SOFR might stay calm while banks are quietly in trouble. That's the basis of the argument for a credit-sensitive supplement.
If Daniel wants to follow the interest rate chain now, what should he actually watch? Give me the dashboard.
First, SOFR itself. The New York Fed publishes it every business day at eight a.Bookmark that page. Second, the effective federal funds rate — published in the Fed's H point fifteen release, which covers selected interest rates. The spread between SOFR and the effective federal funds rate tells you about conditions in the repo market specifically versus the broader interbank market. Third, and this is the modern equivalent of the old LIBOR-OIS spread that screamed danger in two thousand eight — watch the FRA-OIS spread.
Break that down.
FRA is a forward rate agreement — essentially what the market expects three-month LIBOR, or now three-month SOFR, to be in the future. OIS is an overnight index swap rate — what the market expects the compounded overnight rate to be over the same period. The spread between them measures the premium banks charge each other for term funding versus rolling overnight funding. When that spread widens, it means banks are worried about something. They're demanding more compensation for lending over a three-month horizon than for lending overnight. It's the market's fear gauge. During the two thousand eight crisis, the LIBOR-OIS spread blew out to over three hundred fifty basis points. Normally it's around ten.
It's the canary that survived the transition. The specific numbers changed, but the concept — the spread between term and overnight expectations — is still the thing to watch.
And for consumer rates, don't expect SOFR to move your mortgage tomorrow. But if SOFR has been trending up for six months, and the FRA-OIS spread is widening, and the ten-year Treasury yield is rising — that combination means bank funding costs are going up and will eventually hit consumer lending rates. The lags are long but the direction is reliable.
To pull this all together for Daniel's question. Banks borrow from each other because deposits are liabilities, not piles of cash, and the payment system creates daily imbalances that have to be settled in central bank reserves. LIBOR was the price of that borrowing, but it was built on hypotheticals and got manipulated by traders who saw the gap between what was reported and what was real. SOFR replaced it with actual transaction data from the repo market — more honest, but secured by collateral, so it doesn't capture bank credit risk the way LIBOR did. The transition was enormous and messy, and the interest rate chain from SOFR to your credit card is real but slow.
If you're following this from outside the financial industry, the actionable move is to watch SOFR, the effective federal funds rate, and the FRA-OIS spread. That trio gives you a real-time read on the plumbing. When the spread widens, pay attention. When SOFR spikes, something's backing up. These are the numbers that replaced LIBOR — not a perfect substitute, but a genuine improvement built on actual trades rather than a morning survey.
There's still that open question. If SOFR is secured by Treasuries, it's measuring the cost of borrowing against the safest collateral in the world. It's not measuring what it costs an unsecured bank to borrow purely on its name. So during the next banking crisis — whenever that comes — does SOFR actually tell us what we need to know? Or does it stay placid while the unsecured market freezes somewhere out of view?
That's the debate that keeps the wonks up at night. The New York Fed's position is that you can add a spread to SOFR to approximate unsecured bank credit risk — and that's what many contracts now do. But the spread has to be chosen in advance, during calm times, and it might not reflect what actually happens during a panic. There's a push for what's called a credit-sensitive rate — something like Ameribor or Bloomberg's BSBY — that captures unsecured bank borrowing costs directly. But those rates have much thinner transaction volumes. So you're trading depth for relevance.
The old problem in a new shape. The deep market isn't measuring quite the right thing, and the thing you want to measure doesn't trade deeply enough to build a benchmark on.
It's the fundamental tension of financial benchmarks. You want them to be based on real transactions, but the transactions that actually happen in volume might not be the ones that capture the risk you care about. LIBOR solved that by just asking people what they thought — and we saw how that ended.
Nine billion dollars in fines and a global rewrite of two hundred trillion dollars in contracts. Not a great trade.
The LIBOR scandal wasn't just about bad actors. It was about a system that trusted hypotheticals over reality for decades because reality was harder to measure. SOFR is the correction. It's not perfect, but it's built on something solid. And that's the lesson that goes beyond finance. If a number matters, you need to know what's underneath it. Is it a survey, or is it a trillion dollars in actual trades? The answer changes everything.
If this changed how you think about what's actually happening behind your bank account, rate the show and tell a friend. We're at my weird prompts dot com.
Now: Hilbert's daily fun fact.
Hilbert: In the early Renaissance, cheesemakers on São Tomé and Príncipe discovered that the island's volcanic soil produced milk that yielded roughly one kilogram of cheese per fourteen liters — nearly double the yield of continental European dairies at the time, which averaged one kilogram per twenty-five liters. The difference was attributed to an unusually high butterfat content in the milk of goats descended from Portuguese breeds that had adapted to the island's mineral-rich grazing.
...fourteen liters to a kilo. That's a very efficient goat.
This has been My Weird Prompts. I'm Corn.
I'm Herman Poppleberry. We'll see you next time.
