Mark Delaney: Juniper, hey — you have a second before we get going?
Juniper Vale: Yeah, what's up?
Mark Delaney: I was at the ATM last night, just pulling out cash, and I had this moment where I thought — wait, is this money actually in here? Like, is there a number on a screen somewhere that corresponds to a real thing? And I couldn't answer it. I genuinely could not answer it.
Juniper Vale: That's exactly where we're starting today.
Mark Delaney: Because the answer is sort of... no? Uh, not entirely? That's what we're getting into — fractional reserve banking. The thing where your bank holds only a fraction of your deposits and lends the rest out. Like, your full balance is yours on demand, and most of it is already gone.
Juniper Vale: Both true at the same time.
Mark Delaney: Both true at the same time. I assumed there was at least some rule, some floor, that commercial banks had to keep a certain amount on hand. And there was. Until March 2020, when the Federal Reserve dropped the reserve requirement to zero percent.
Juniper Vale: Zero. Not low. Zero.
Mark Delaney: Zero. American commercial banks face no mandated minimum reserve ratio right now. And most people have no idea that happened.
Juniper Vale: Okay, I want to make that concrete, because it's easy for that number to just kind of slide past. Think about a restaurant owner who wires $220,000 into her business account on a Friday afternoon. By Monday morning, most of that balance is a number on a screen. The bank has already deployed the bulk of it — into someone else's mortgage, a business loan, gone. And there's no rule that says how much has to stay.
Mark Delaney: So what actually holds it together, then? If it's not a rule — what's the circuit-breaker?
Juniper Vale: The circuit-breaker is a coat-check room. Think about that — you hand your coat over, you get a ticket, you go to dinner. What you don't know is that the coat-check is quietly renting your coat to someone else while you're eating. It works perfectly. Every single night. As long as not everyone tries to leave at the same time.
Mark Delaney: And the whole thing just... runs on that bet.
Juniper Vale: That's the whole system. Commercial banks accept your deposit, hold a fraction as reserves — either in vault cash or parked at the Federal Reserve — and lend the rest. The core idea is just: not everyone will ask for their coat tonight.
Mark Delaney: And this isn't new, right? Like, this goes way back.
Juniper Vale: Medieval goldsmiths. You'd deposit your gold with a goldsmith for safekeeping, he'd give you a receipt, and then — and this is the discovery — he'd notice that most people just... didn't come back for it. Not all at once, anyway. So he'd lend some of it out, collect interest, and as long as withdrawals stayed predictable, nobody knew the difference.
Mark Delaney: Wait, so the entire system we have now — the Federal Reserve, commercial banks, all of it — traces back to some guy in the 1300s going 'huh, nobody's asking for this gold back, I wonder...'
Juniper Vale: Basically, yeah. The foundational insight is that simple. And the textbook version of this — the financial intermediation view — says banks are just pass-throughs. They collect savings, route them to borrowers. You're the saver, someone else is the borrower, the bank is the pipe in the middle.
Mark Delaney: Which sounds clean. Like a utility.
Juniper Vale: It does. But here's where it breaks — if that were the whole story, dropping the reserve requirement to zero should have snapped something. Like, that's the rule that sets the ceiling on how much lending can chain out from a single deposit. The money multiplier only works if there's a binding floor. The Federal Reserve removed it in March 2020, and nothing broke. So either the model was wrong, or something else entirely is doing the load-bearing work.
Mark Delaney: If the rule isn't actually — wait, no — what I'm trying to get at is, the system kept running. Which means the thing holding it up isn't the regulation. It's something else.
Juniper Vale: It's the assumption that depositors won't panic. That's it. Not a mathematical constraint. Not a legal floor. A behavioral one. The system isn't built on the probability that you won't withdraw — it's built on the belief that enough of us trust it enough not to.
Mark Delaney: That's... kind of a terrifying way to run a financial system, man.
Juniper Vale: Terrifying — but I want to push on *why* it's terrifying, because I think the textbook story actually undersells how weird this is. The financial intermediation view, the classic one, says banks are just moving existing money around. But that's not what Richard Werner found.
Mark Delaney: Werner — he's the one who ran the empirical test, right? Like, actually went into a bank?
Juniper Vale: First direct bank-level test distinguishing all three banking theories. And what he found — this is the part that messes with the clean version — when a bank extends a loan, it doesn't move money from one account to another. It creates a new deposit on its own balance sheet, right at that moment. The purchasing power didn't exist before. It does now.
Mark Delaney: Hold on. So it's not recycling savings that already exist somewhere.
Juniper Vale: No. The credit creation theory says: the money comes into being at the moment of lending. Not downstream through some multiplier chain — simultaneously, on the bank's own books.
Mark Delaney: The money multiplier story, the $1,000 becomes $10,000 through successive rounds, that's actually describing something that doesn't quite happen the way we think?
Juniper Vale: The multiplier model assumes a sequential chain — deposit, lend, re-deposit, lend again. But if each individual commercial bank is creating a new deposit at the moment it lends, that chain isn't the mechanism. It's more like... each loan is its own origin event.
Mark Delaney: And the IMF weighed in on this? Like, the actual IMF?
Juniper Vale: 2016 analysis. They came down on the credit-creation side over the traditional intermediation story. Which is not a small thing — that's the institution that basically advises governments on how banking works, saying the textbook framing is off.
Mark Delaney: In 2008, banks had reserves. They weren't empty. And they just... stopped lending. The multiplier froze. Not because of some mathematical ceiling — confidence broke, and the whole mechanism just sat there.
Juniper Vale: Wait — that's the thing, isn't it. If the multiplier were purely mechanical, reserves plus ratio equals lending capacity, it should have kept running in 2008. But it didn't.
Mark Delaney: Which means the credit creation engine — the part where commercial banks conjure new deposits — is downstream of psychology. You have to *want* to lend. Borrowers have to *want* to borrow. If that belief breaks, the math is just sitting there doing nothing.
Juniper Vale: And that's actually going to matter a lot when we get to what happens when enough depositors lose that belief at the same time — because the run mechanics and what the lender of last resort can actually do about it, the unsettling part is what happens there.
Mark Delaney: So the circuit-breaker we haven't talked about yet is the one that has to run on confidence too. That's — yeah, I want to get there.
Juniper Vale: And that's exactly the thread — because what the run mechanics reveal is something that doesn't get said plainly enough. Withdrawing your money when you think the bank is in trouble is not irrational. It's correct. It's the right call.
Mark Delaney: Wait — it's the right call?
Juniper Vale: Individually? Yes. You fear the bank can't cover everyone. You move first, you're safe. That is the individually correct move. But here's what makes it genuinely unsettling — your rational choice is exactly what creates the catastrophe you were afraid of.
Mark Delaney: Because everyone else is running the same logic at the same time.
Juniper Vale: Sequential withdrawals drain the reserves. The bank hits the floor, has to start selling assets fast — and distressed asset sales at a discount can push a bank that was actually solvent into insolvency. Not because it was insolvent. Because everyone acted as if it was.
Mark Delaney: The fear makes the thing real. That's — man, that's dark. So what actually stops it?
Juniper Vale: The lender of last resort. The central bank — the Federal Reserve in the U.S. — steps in with emergency liquidity. Not to a bank that made bad loans and is genuinely broke, but to a bank that's illiquid: has good assets, just can't convert them fast enough to meet a sudden withdrawal surge.
Mark Delaney: Okay but — uh, doesn't that mean banks know the cavalry's coming? Like, if you know someone's going to bail you out, do you actually manage risk the same way?
Juniper Vale: That's the moral hazard, and it's the exact price of the backstop. The lender-of-last-resort function stabilizes the system — and by doing that, it quietly removes some of the market discipline that would otherwise keep banks from taking on too much risk.
Mark Delaney: You buy stability with recklessness. Or, like — you buy it with the *permission* for recklessness.
Juniper Vale: And someone actually formally proved the tradeoff. Heon Lee built a dynamic monetary model — and I mean genuinely formal, mathematical — showing that lower reserve requirements raise steady-state consumption. Real economic output goes up. But simultaneously the system becomes more prone to endogenous cyclical instability, chaotic dynamics. It's not a feeling. It's in the model.
Mark Delaney: So more efficient and more fragile at the same time, and you can't pick just one.
Juniper Vale: The Threadneedle simulation — it's a multi-agent model built on actual double-entry bookkeeping — runs fractional reserve banking under different regulatory assumptions and keeps landing in the same place. No clean resolution. Every setting that improves efficiency degrades stability somewhere.
Mark Delaney: Huh. So it's not that we haven't figured out the right ratio yet. It's that there isn't one.
Juniper Vale: Right — but the part that doesn't fit cleanly into even that model is what digital banking does to the timeline. Physical queues used to be the binding constraint on a run. You had to physically show up. That friction bought time — for the central bank to notice, to act. Now withdrawals are instantaneous and borderless. Confidence can collapse in the time it takes a tweet to spread.
Mark Delaney: So the lender of last resort is running a slower clock than the panic is.
Juniper Vale: That compression — the gap between how fast belief can break and how fast the Federal Reserve can respond — I think that's the deepest fragility the system carries right now. Not the reserve ratio. Not even the moral hazard. The speed.
Mark Delaney: And every fix I can think of just... moves the problem somewhere else. Full-reserve banking — okay, you require commercial banks to hold 100% against demand deposits. No more runs. But you've also just killed the credit-creation engine. Mortgages, small business loans, the whole expansion mechanism — that's gone. You've bought stability by unplugging the thing that funds growth.
Juniper Vale: Central Bank Digital Currency proposals do the same thing in a different coat. You let people hold accounts directly at the Federal Reserve, no commercial bank in the middle — and yes, the intermediation risk disappears. But now the government is the lender for the entire economy. Every mortgage, every business loan. That's not obviously better. It's just a different address for the tradeoff.
Mark Delaney: Islamic finance, narrow banking — uh, same story, right? You relocate where the fragility lives, you don't actually get rid of it. I don't think there's a version of this where you keep the private credit-creation engine running and also make it fully stable. I've been trying to find one and I keep coming up empty.
Juniper Vale: The thing I'm still turning over — the Federal Reserve proved in March 2020 that the system doesn't actually require reserves. Zero percent. And it held. So what it requires is that enough people believe it doesn't need them. That's not a mechanical property. That's a belief.
Mark Delaney: That's not engineering. That's collective psychology wearing a spreadsheet's clothes.
Juniper Vale: You know what's funny — you started this whole thing at an ATM wondering if the money was actually in there. And the answer is still sort of no. It's just that now we know the thing holding it together isn't a rule or a ratio. It's everyone quietly agreeing not to ask at the same time. I think that's where we are.
