Mark Delaney: Juniper, hey — you see the news this week, any of it, or were you just deep in spreadsheets again?
Juniper Vale: Mostly spreadsheets, honestly. But I did surface long enough to think about today's topic, which — okay, I have a number for you, and I want your gut reaction before we get into any of it.
Mark Delaney: Hit me.
Juniper Vale: Two years. Over two full years of the yield curve inverted — the longest stretch on record — and the U.S. economy grew 2.9% in 2023, over 3% in 2024, and the National Bureau of Economic Research never called a recession. Not once.
Mark Delaney: Wait, seriously? Over two years and — nothing?
Juniper Vale: Nothing. Which matters because the yield curve inversion is basically Wall Street's most cited recession signal. You know — the yield curve is just a graph of what U.S. Treasuries are paying across different time horizons, short to long. When it flips — when short-term bonds are paying more than long-term ones — that's the inversion. And historically that flip has come before every single recession since 1960.
Mark Delaney: Every single one. So it fires, uh, every time — and then this time it just... sat there for two years and the economy kind of shrugged.
Juniper Vale: That's the puzzle. Did the signal break, or did something change about the world it was measuring? That's what we're trying to figure out today.
Mark Delaney: And before we can even ask why it failed, I don't totally get why it works in the first place. Like what is the curve actually telling you?
Juniper Vale: Okay, the core idea is actually really simple. Imagine you're lending a friend money — a week versus ten years. You'd charge more for ten years. More can go wrong. That's just intuition.
Mark Delaney: Yeah, obviously. Ten years is riskier.
Juniper Vale: Right — and that extra charge you're demanding, that's what economists call the term premium. The compensation for sitting in a long bond while inflation could move, rates could shift, anything could happen. That premium is why the yield curve normally slopes upward. It's not random, it's baked in.
Mark Delaney: So when it flips — when short rates are higher than long ones — something has overwhelmed that premium.
Juniper Vale: Exactly. And that something is expectations. There's a framework called expectations theory — the idea that a ten-year yield is basically the market's best guess at what short-term rates will average over the next decade. So if investors think the Federal Reserve is going to cut rates sharply, future short rates look really low. That drags long yields down below today's short yields. Inversion.
Mark Delaney: Wait — so it's not some technical glitch, it's just... millions of investors voting with money on what they think the Fed does next.
Juniper Vale: That's it. Ben Bernanke actually spelled this out — long rates embed both the expectations component and the term premium, and when the expectations piece overwhelms the premium, you get inversion. The Federal Reserve Bank of New York built their whole recession-probability model on exactly this mechanism. It's not mysterious. But — and this is the part that doesn't let me fully relax about it — if the Federal Reserve itself can suppress where people expect rates to go, it can kind of... move the goalposts on what the signal means.
Mark Delaney: So the signal is sound but the thing it's measuring — Fed expectations — that's the variable that can get, uh, manipulated isn't the right word, but shaped.
Juniper Vale: Shaped — yeah, that's actually the better word. And that's where the track record gets interesting, because the record itself is genuinely impressive before you start poking at it. The 3-month/10-year spread — not 2s/10s, the other one — has preceded every single U.S. recession since 1960. Every one, per NBER dating.
Mark Delaney: Every one since 1960. That's, uh — that's not nothing.
Juniper Vale: It's not nothing. The 2s/10s spread, which is the one you actually see in headlines, has only preceded seven of the last eight recessions since 1968. It missed one. And the San Francisco Fed came out and said, look, the 3-month/10-year is the more reliable measure. So there's already a disagreement inside the Federal Reserve system about which version of the signal you should even be watching.
Mark Delaney: Hold on — so the most-cited spread might be the less accurate one?
Juniper Vale: That's what the San Francisco Fed is basically arguing, yeah.
Mark Delaney: Okay and then — wait, the lead time thing too. Because even if the signal fires correctly, I keep thinking, uh, 7 to 24 months is a pretty wide window. Like that's the difference between selling your stocks today and sitting on that decision for two years.
Juniper Vale: Fifteen months on average, but yeah — the range is almost useless for actual timing. And it gets crackier when you go outside the U.S. The UK and Canada have both had inversions that weren't followed by a domestic recession. So the mechanism that seems structural here just... didn't travel.
Mark Delaney: That's kinda the problem. Durable in the U.S., fine — but if it produces false alarms in the UK and Canada, what does durable actually mean?
Juniper Vale: And that question gets a lot harder when you factor in what quantitative easing does to term premiums — because there's a version of this where the Federal Reserve can mechanically cause an inversion that has nothing to do with real recession expectations, and the 2022-to-2024 stretch might be exactly that case. We'll get there.
Mark Delaney: Yeah and that's — okay, mechanically, how does QE actually do that? Like, I get that the Fed buys bonds, but how does buying bonds flip a signal?
Juniper Vale: So when the Federal Reserve buys long-duration Treasuries at scale — which is what quantitative easing is — it pulls those bonds out of the market. Less supply. Prices go up, yields go down. That directly compresses the term premium. The extra compensation investors normally demand for holding long bonds? The Fed just... absorbed it.
Mark Delaney: So the curve can invert just because the Fed is hoovering up long bonds. Not because anyone actually thinks a recession is coming.
Juniper Vale: That's exactly the problem. And the Boston Fed named it — their researchers looked at 2019 specifically and said the inversion there likely overstated recession probability because monetary policy was already so accommodative. The federal funds rate was well below its neutral level. Part of what you were seeing wasn't fear of recession, it was the policy stance itself bending the curve.
Mark Delaney: Hold on — 2019, not 2022. That's a different inversion entirely.
Juniper Vale: Different inversion, same distortion. And then on top of that, you also get flight to quality — investors who are nervous buying long Treasuries as safe havens, pushing yields down further. So now you've got two forces compressing long yields mechanically, neither one of which is a pure read on growth expectations. The ECB actually published research on this — acknowledged that unconventional monetary policy complicates the whole interpretation of U.S. curve inversion as a recession signal. It's not just a domestic debate.
Mark Delaney: Wait, the European Central Bank weighed in on what the U.S. curve means? That's — uh, I mean, that tells you this is a bigger conversation than just Fed-watchers arguing.
Juniper Vale: It really does. And think about — late 2022, someone watching their retirement account, sees the inversion headline, the longest on record, and they're trying to decide whether to move to cash. The signal guiding that decision was partly manufactured by the Federal Reserve buying bonds. That's not the signal breaking, that's the signal measuring something real — but something that includes the Fed's own footprint. You can't separate them out.
Mark Delaney: So it's less broken, more... the Fed became part of what it's measuring. Which is a genuinely unsettling thing to realize about something people treat as gospel.
Juniper Vale: And that's actually the part that matters, I think. Not whether it broke — but whether the question we've been asking it is still the right one. Because if the 2s/10s spread and the 3-month/10-year spread are both still measuring real investor anxiety — real fear about where the Federal Reserve is headed — then the signal isn't wrong. It's just stopped being a countdown clock.
Mark Delaney: Yeah, no — that's it. Maybe we've been asking it the wrong question this whole time. 'When does the recession start' when it's actually more like... 'how scared are bond investors right now.' Which is useful, uh, just useful in a totally different way than we thought.
Juniper Vale: The real test is the next inversion — whenever it comes — under conditions where the Fed isn't suppressing term premiums. That's when we find out if the Boston Fed was right that this was a policy artifact, or if something more structural has shifted. We genuinely don't know yet.
Mark Delaney: Which is an uncomfortable place to leave it.
Juniper Vale: It is. But I'd rather sit with that honestly than wrap it up cleaner than it deserves. Thanks for going down this one with me.
