Finn Brooks: Okay I have to tell you about something that happened to me Tuesday — I was explaining to my roommate why his money in a savings account is basically being lent to strangers, and he looked at me like I had just told him his house was haunted.
Clara Bennett: How far did you get before he stopped trusting banks entirely?
Finn Brooks: I got to the part where I tried to explain the reserve requirement — ten percent reserves, ninety percent lent out, money multiplier — and then I realized mid-sentence that I actually had no idea if that's still true. Because didn't the Federal Reserve just... get rid of that?
Clara Bennett: March 2020. Reserve requirement ratio: zero percent. That's the fact I want to start with today, because — now, on its face that sounds alarming. But the more interesting thing is what it signals. It's not that the Fed decided banks could suddenly lend unlimited amounts. It's that the ratio had already stopped being the operative limit. They removed a floor that nothing was resting on.
Finn Brooks: Wait — so the formal requirement was just decorative at that point?
Clara Bennett: Effectively. Which means the story most people know — and the story we still teach — the money multiplier, the elegant ratio machine where a ten percent reserve requirement produces a multiplier of ten — that story had been running on autopilot, describing a constraint that had already gone soft. And in March 2020, the Federal Reserve made it official.
Finn Brooks: That is — okay that's actually kind of a remarkable thing to admit publicly. Like, the textbook explanation and the real system had just quietly diverged and nobody updated the textbook.
Clara Bennett: That divergence is what this episode is about. The system we're describing is fractional reserve banking — commercial banks hold a fraction of deposits as reserves, lend out the rest, and in doing so expand the money supply beyond what the monetary base alone would allow. That basic structure is real and it's still operating. The question I want us to work through is: if the reserve requirement isn't actually what governs the system anymore, what is? And does the underlying fragility — the bank run problem — change at all when the constraint shifts?
Finn Brooks: And my roommate, for the record, is now keeping his savings in cash under his mattress.
Clara Bennett: Which is technically full-reserve banking at the personal level — we'll get to why that doesn't scale.
Finn Brooks: Okay I need to know everything.
Clara Bennett: Before we get to why that doesn't scale — the middleman picture. Because I want to make sure we've actually said it plainly. A bank is just a very trusted middleman: you deposit $100, it lends $90 to your neighbor, your neighbor deposits that $90 somewhere, and now there's effectively $190 in the system from your original $100. The bank never had both dollars at the same time. It never needed to.
Finn Brooks: That's the part that made my roommate's face go blank, actually. Like — wait, $190 exists but only $100 was ever real?
Clara Bennett: Right. And the textbook gives that a formula. Ten percent reserve ratio, multiplier of ten — so that same $100 deposit theoretically supports up to $1,000 in total deposits once you run through enough rounds of lending and redepositing. That's the money multiplier in its cleanest form.
Finn Brooks: Okay and I actually — I want to stress-test this, because that math feels almost too satisfying? Like, it closes so neatly. One number in, one number out.
Clara Bennett: That's exactly the tell. In practice, the actual multiplier never reaches that ceiling. Two things pull it down — banks voluntarily hold excess reserves beyond what any rule requires, and depositors keep some cash outside the banking system entirely. Money sitting in a wallet doesn't redeposit. So every round leaks a little.
Finn Brooks: Hold on — voluntarily? Banks choose to hold more than they have to?
Clara Bennett: Consistently. And this isn't a new behavior — it predates the Federal Reserve's March 2020 move to zero. Banks were already sitting on excess reserves well before that. Which means the multiplier was already falling short of its theoretical maximum before the formal requirement was even removed.
Finn Brooks: So the model was overstating credit creation — like, not a little, but structurally, every time?
Clara Bennett: Every time. And here's what that makes awkward: the money multiplier is still the dominant way we teach this. High school, intro economics, policy conversations — it's the frame. But it's a model that describes a ceiling banks were never actually hitting, constrained by a ratio the Federal Reserve eventually just... set to zero.
Finn Brooks: That's — okay that gap between the teaching framework and actual credit creation, that's not a rounding error. That's the entire mechanism being wrong.
Clara Bennett: The mechanism isn't wrong, exactly — it's incomplete. The real question is what was it actually measuring, if not the binding constraint on lending? And that's where the model stops being a description of how banks behave and becomes more like a... pedagogical artifact.
Finn Brooks: A pedagogical artifact. So we built an entire intuition for how money works around a formula that was already running ahead of reality — and nobody pulled the fire alarm until the Federal Reserve literally eliminated the number it was built on.
Clara Bennett: And that's actually the thread I want to pull — because if the reserve ratio wasn't the real constraint, what is? The answer is Basel III. Which sounds like a bureaucratic acronym, but in practice it's the thing actually doing the work the reserve ratio was supposed to do.
Finn Brooks: Wait, like — doing the work how exactly?
Clara Bennett: Three dimensions. Minimum capital ratios — the bank has to hold enough equity to absorb losses. Liquidity coverage ratios — enough liquid assets to survive a 30-day stress period. And leverage ratios — a ceiling on how much total exposure you can carry relative to your capital base. None of those are in the money multiplier formula. Not even close.
Finn Brooks: So a bank hits a lending ceiling not because it runs short on reserves — it runs short on capital cushion first.
Clara Bennett: Exactly. And that's — I mean, that's a structurally different kind of brake. A reserve requirement scales with deposits. A capital ratio scales with risk-weighted assets. Those are not the same animal. The classic money multiplier formula entirely ignores that distinction.
Finn Brooks: Okay but here's what I keep snagging on — if Basel III is now the real constraint, why did loosening the old reserve requirement even matter? Like, why not just shrug?
Clara Bennett: Because there's a researcher — Heon Lee — whose modeling actually answers that. Lower reserve requirements correlate with greater endogenous economic volatility. Not immediately, not obviously, but structurally. Loosen one scaffold without building an equivalent replacement and the system gets more unstable, not the same.
Finn Brooks: Wait — greater volatility from lower reserve requirements? That feels backwards. I'd have assumed loosening a constraint just... frees up credit.
Clara Bennett: It does free up credit. That's the problem. More credit creation, less friction — and no, wait, the issue is the credit creation was already happening beyond the reserve constraint. So removing it doesn't add meaningful lending capacity, it just removes a signal. And signals, it turns out, matter even when the underlying rule doesn't bind.
Finn Brooks: So we've been layering new scaffolding — Basel III on top of a framework whose original scaffold the Federal Reserve literally zeroed out in March 2020. And the Bank of England in 1694, the Swedish Riksbank going back to 1668 — every single generation has been doing that same thing. Adding a new institutional layer to contain fragility that the last layer couldn't quite hold.
Clara Bennett: And still needing another one. That's the historical irony that doesn't get said plainly enough.
Finn Brooks: Which — honestly, that's what makes the next part feel kind of ominous, because the structural thing underneath all those scaffolds, the maturity mismatch, that's still completely intact, and digital technology changed something qualitative about how fast it can detonate.
Clara Bennett: That's exactly where we're going — and the numbers from 2023 make it very concrete.
Finn Brooks: Okay but the maturity mismatch — that's the thing underneath all of it, and I want to say it plainly because I don't think it fully landed for me until recently. Banks borrow short and lend long. Your deposit is liquid — you can pull it Tuesday morning. A 30-year mortgage the bank issued with that deposit is completely illiquid. The bank cannot call that mortgage back. Ever.
Clara Bennett: And that's not a flaw someone could engineer out. That mismatch is how the system generates credit in the first place.
Finn Brooks: Which means — wait, this is the part that genuinely unsettles me — Basel III doesn't touch it. Like, capital ratios and liquidity coverage ratios are doing real work, but they're raising the threshold for panic, not eliminating the structural gap that panic exploits.
Clara Bennett: Exactly. They raise the confidence threshold. They don't remove what's underneath.
Finn Brooks: Silicon Valley Bank. Spring 2023. A venture capital portfolio manager sees a news alert — SVB's bond portfolio is down $15 billion. She picks up her phone, transfers $5 million to JPMorgan. Before her coffee's cold. Her colleague gets a transaction notification, does the same. By end of day: $42 billion in withdrawal requests. One day.
Clara Bennett: Forty-two billion. In a single day.
Finn Brooks: And SVB was — this is the part — technically solvent. Capital ratios intact. Not a zombie bank propped up on bad loans. It evaporated in roughly 48 hours because confidence failed, not because the underlying math failed.
Clara Bennett: Now, the 1930-31 banking panics worked through exactly the same mechanism — depositor fear cascading across institutions, collapsing banks that were arguably solvent. That part isn't new. What's new is the speed. A 1930s bank run took days to develop. People had to physically queue. That friction was — I mean, it was terrible for individuals, but it was also involuntary circuit breaker.
Finn Brooks: And now that friction is gone.
Clara Bennett: Completely. Digital transfers, social media — a tweet reaches a depositor before any regulator can respond. The psychological trigger is qualitatively more dangerous now, not just faster. The information that used to take a day to spread takes four minutes.
Finn Brooks: So the maturity mismatch was always the loaded gun. Digital technology just — it removed the safety. No, actually, it's worse than that. It automated pulling the trigger.
Clara Bennett: That's the irreducible danger. Deposit insurance, lender-of-last-resort functions — they're real stabilizers. But they work by shoring up belief. The moment belief moves faster than any institution can respond, the structure is exposed.
Finn Brooks: And that's what SVB proved. Not that regulation failed — that the gap between liquid liabilities and illiquid assets is always there, waiting, and now any depositor with a smartphone can act on it before the person next to them even knows there's a problem.
Clara Bennett: And the proposed fixes — I mean, philosophically it's strange. Irving Fisher in 1936 said: require reserves equal to 100% of deposits. Full reserve banking. No private money creation, no maturity mismatch, no bank runs. Clean.
Finn Brooks: Wait — 1936, that's the middle of the Great Depression. He watched those 1930-31 banking panics cascade and his answer was just, eliminate the gap entirely?
Clara Bennett: That was the logic. But the critique isn't that it fails to stop bank runs — it probably does stop them. The critique is that it shifts the problem. If banks can only lend what they actually hold in long-term deposits, credit becomes rigid. The timing mismatch moves from inside the bank to somewhere in the broader economy — intertemporal discoordination, the term is. And then there's the political reality, which is that no country has ever actually implemented it at scale. The feasibility problem is nearly as large as the theoretical one.
Finn Brooks: So it relocates the tension. It doesn't dissolve it.
Clara Bennett: Same pattern with CBDCs — digital sovereign money issued directly by a central bank, cutting out the commercial bank deposit step entirely. No private credit creation. In theory, that solves the confidence problem because the liability sits with the central bank, not a fragile intermediary. But now ask who decides where credit flows. That power — now, that's concentrated in a sovereign institution with no market signal telling it what to fund.
Finn Brooks: That's — yeah, trading one fragility for a completely different category of risk. The bank run disappears and you get something closer to a planning problem.
Clara Bennett: Which brings me back to what I actually can't stop thinking about. Not whether the reserve ratio model was wrong — it was. But whether the institutional scaffolding can be redesigned fast enough to keep pace with the speed at which digital technology lets confidence evaporate. Deposit insurance, central bank backstops, capital rules — those were built for a world where a bank run took days. SVB took 48 hours. The next one might take four.
Finn Brooks: And here's the part I keep sitting with — medieval goldsmiths didn't design any of this. They just noticed that not everyone came back for their gold at once. That behavioral observation, that accidental gap, that's the foundation the entire modern financial architecture is built on. Irving Fisher, Basel III, CBDCs — all of it is downstream of a goldsmith in the 12th century going, huh, nobody's actually coming back for this.
Clara Bennett: Your roommate's money under the mattress is starting to look coherent. Not scalable — but coherent.
