Lila Soto: Iris, long week — but I cannot stop thinking about this one moment. Can I just set a scene?
Lila Soto: Columbus, Ohio. Retired schoolteacher. She's got $240,000 in a Wells Fargo checking account — money she spent thirty years saving — and it's earning nothing. Then she notices a poster at her credit union: 4.8% on a money-market fund. And she moves the whole thing.
Iris Holm: How much does that cost Wells Fargo?
Lila Soto: That's exactly what I wanted to get at. Because here's what's wild — the same quarter she does this, Wells Fargo's stock pops 6.1% on earnings. The market is cheering. But non-interest income is actually down 1.4% year-over-year.
Iris Holm: The market read the headline. Net interest income looked good. It missed the offset.
Lila Soto: So — I mean, for people who have never thought about this stuff — what is a bank actually doing to make money? Like, what's the basic engine?
Iris Holm: One sentence: a bank borrows cheap from depositors and lends expensive to borrowers, and the gap — the net interest margin — is the profit. That's it. JPMorgan, Citigroup, Bank of America all grew that line by 6, 7, 8 percent in Q3. The question is whether that gap is actually as wide as the earnings calls suggest, or whether the teacher in Columbus is already narrowing it.
Lila Soto: But that's exactly when the floor drops out, right? Because just as those Q3 numbers land — the October CPI report comes in. Up 0.4%, seasonally adjusted. More than anyone forecast.
Iris Holm: Goolsbee calls it 'borderline disturbing.' That's the Chicago Fed president. That's not throwaway language.
Lila Soto: So more hikes are coming?
Iris Holm: Christopher Waller put a half-point December hike on the table if core CPI came in around 0.3. It did. So — look, the Federal Reserve is sitting at 3.75 to 4.00 percent, holding, but actively debating whether to accelerate. That's the pivot point.
Lila Soto: Okay but — I mean, isn't that good for banks? Higher rates, wider spreads?
Iris Holm: That's the cruel part. Every additional hike tells depositors that money-market yields will stay elevated — which signals them to move faster. The same rate signal that might widen a bank's lending spread is also the bell that sends the teacher in Columbus to that credit union poster. Deposit migration accelerates. Funding costs go up. And the net interest margin gain you just reported? It starts getting eaten before the next quarter closes.
Lila Soto: So the hike is simultaneously the win and the thing that — yeah, unwinds the win.
Iris Holm: Deposit migration isn't a one-time event. It's a ratchet. Once customers shift from noninterest-bearing checking into CDs or money-market funds, they don't shift back when the next earnings call looks pretty. Banks with weak deposit franchises — ones that don't have sticky, low-cost core deposits — they're already locked into paying more. And if Waller's half-point lands in December, the FDIC data on community-bank NIM at 3.81 versus the industry 3.32 — that gap probably closes fast, just in the wrong direction.
Lila Soto: So Q3 2026 might literally be the peak — and nobody's reading it that way yet.
Iris Holm: And that gap closing fast — that's where the ranking problem actually lives. Because the intuition is 'JPMorgan wins, it's big, it's diversified.' But the FDIC data doesn't support a size thesis. Community banks posted 3.81% NIM versus 3.32% industry-wide in Q2 2026. Bigger isn't the variable. Asset sensitivity is.
Lila Soto: Wait — community banks are actually outperforming JPMorgan on spread?
Iris Holm: On NIM, yes. Because a lot of them hold short-duration floating-rate loans — small business credit, agricultural lines — that reprice almost immediately when the Fed moves. Their assets reset faster than their liabilities. That's asset sensitivity. The spread widens before deposits have time to migrate.
Lila Soto: Okay but — I mean, isn't the flip side that those same depositors, the ones at a small community bank, are maybe more likely to notice a 4.8% CD rate and move? Like, they're in a tighter community, word travels. So the advantage maybe... collapses faster?
Iris Holm: That's the whole race. Loan repricing speed versus deposit cost acceleration. Imagine a produce distributor in Fresno — $400,000 floating-rate line of credit, resets every quarter. His bank captures that immediately. But if his neighbor moves savings into a Treasury fund in the same week — same bank, net effect narrows. No clean winner. The outcome is literally a timing problem.
Lila Soto: And that's why JPMorgan's scale doesn't just... automatically solve it.
Iris Holm: Right — what JPMorgan has is a low-cost deposit franchise. Millions of checking accounts, noninterest-bearing, people who just don't move. Funding costs rise slowly for them. But a regional bank with concentrated commercial deposits? Those clients are sophisticated. They move on a Tuesday afternoon. So the question isn't large versus small — it's franchise stickiness versus asset sensitivity, and there's genuinely no definitive ranking over a prolonged higher-rate cycle.
Lila Soto: Huh. And the stickiness question — I kind of want to pull that thread more, because there's something underneath it about whether we even measured deposit inertia correctly last time around that I think changes all of this.
Iris Holm: That's exactly where this gets uncomfortable. And we'll get there.
Lila Soto: And 'measured correctly last time' — that's the part that keeps nagging at me. Because the 2022–23 cycle was supposed to teach us something about deposit inertia, and I'm not sure we actually learned it.
Iris Holm: We didn't. The lesson from that cycle is that deposit stickiness was systematically overstated. Banks modeled customers as slow-moving. They weren't — they just needed enough rate differential to make the friction worth it. Once that threshold crossed, they moved. Fast.
Lila Soto: So we're pricing bank stocks today as if the lesson landed — but maybe it didn't?
Iris Holm: That's exactly the bet the market is making. And look — the post-2008 era is the bookend nobody's invoking. The Fed held rates near zero for years. Years. NIM stayed compressed for years. It wasn't a quarter, it wasn't two. The rate environment reshaped bank profitability durably, in one direction. The same durability cuts the other way now. If rates stay elevated, deposit competition doesn't peak — it compounds. Every month that passes, more customers figure out the math. And if the Fed actually pivots in mid-2027, banks could be sitting on higher funding costs with no repricing upside left. That's the trap.
Lila Soto: Oh — so duration is the knife that cuts both ways.
Iris Holm: Right. And the Romer and Romer research is relevant here — their finding is that the Fed's commitment to reducing inflation at the outset is what determines whether disinflation actually holds. Weak commitment, small short-lived decline. But Goolsbee calling October 'borderline disturbing' while the Fed holds in October and Waller floats a half-point in December — that's not strong commitment, that's mixed signals. Which means the rate path is genuinely unresolved. Banks can't model duration if the Fed can't commit to a direction.
Lila Soto: Hm. So the uncertainty is almost worse than a clear hike.
Iris Holm: Picture a loan officer in Boise, December 2026. She's setting deposit rates for Q1. Does she price for one more hike or a plateau? She doesn't know. So she underprice — and if Waller's half-point lands, she's already behind. That's the Romer problem made concrete. The commitment signal is what anchors the model, and right now it's blinking.
Lila Soto: And she's the person who finds out in Q1 2027 earnings that Q3 2026 was actually the peak.
Iris Holm: And she finds out the same way the teacher in Columbus already did. The Q1 2027 earnings call runs the numbers. JPMorgan's 6% NII growth, Bank of America's 8% — those might literally be the ceiling. Not the floor.
Lila Soto: The Q3 report doesn't know the money moved. That's the part that just... sits with me.
Iris Holm: The deposit is gone. The balance sheet hasn't caught it yet. If Waller's December hike lands and betas spike in Q1 — that compression shows up in earnings calls that everyone will read as a surprise. It won't be a surprise. The teacher already told them.
Lila Soto: Yeah. I think that's the one thing I actually walked away holding — that the signal was already in the behavior, not in the report.
Iris Holm: Good place to stop. Thanks for the scene from Columbus. It did the work.