Treasury Yields Are Rising: What That Actually Does to Stocks - and to Which Ones
📈 Treasury Yields Are Rising: What That Actually Does to Stocks — and to Which Ones
Welcome back, everyone! 📊 I'm your Quant Analyst, filtering out market noise using data, statistical modeling, and systematic insights. 👩💻✨
“Rates up, stocks down” is the most repeated sentence in market commentary and one of the least useful, because it answers neither of the two questions that decide what actually happens to a portfolio: which part of the curve moved, and what multiple you own.
So let me answer both with arithmetic. My claim: this year’s move is a bear flattener — the short end has risen far more than the long end — and that is a materially different signal than the fiscal-panic steepener people usually fear. The damage is real, but it is concentrated by valuation, and I can show you exactly how concentrated.
▲ Every rate discussion eventually reduces to one line and one denominator
📌 First, the reported numbers
All figures are daily constant-maturity Treasury yields from the Federal Reserve’s H.15 release, as published on FRED, for August 27, 2026: 2-year 4.20%, 5-year 4.38%, 10-year 4.67%, 30-year 5.19%. Year to date (from January 2, 2026) that is +73bp on the 2-year, +64bp on the 5-year, +48bp on the 10-year and +33bp on the 30-year. The 10-year’s range over the past year is 3.97%–4.75%, so it currently sits near the top of it.
Source: Federal Reserve H.15 via FRED, series DGS10, DGS2, DGS30.
1. Read the Curve Before Reading the Headline
The single most useful thing in that data box is not the level. It is the gradient: the shorter the maturity, the bigger the rise.
Which End of the Curve Actually Moved
*Actual reported yields, not a model. Change from Jan 2 to Aug 27, 2026, Federal Reserve H.15 via FRED. The downward slope left-to-right is the whole story of this section.
That shape has a name: a bear flattener — yields rising, with the front end rising fastest. It matters because the two ends of the curve are driven by different things, and therefore carry different messages.
- The front end is a Fed-expectations instrument. A 2-year yield is roughly the market’s average expected policy rate over two years. When it jumps 73bp, the market has repriced what it thinks the Fed will do — in this case, toward tighter-for-longer rather than the cuts that were priced in January.
- The long end is a growth, inflation and term-premium instrument. A 30-year yield says more about the market’s long-run inflation view and how much it demands to be paid for duration risk. It rose only 33bp.
So the honest translation of this year’s move is: the market has changed its mind about the Fed, and has barely changed its mind about the long run. That is a very different animal from the scenario people usually have in mind when they get nervous about rates — a bear steepener, where the long end leads because investors are demanding more compensation for inflation or fiscal risk. That version questions the denominator of every asset. This version mostly re-prices the next two years of policy.
It also connects to something I wrote about earlier this month. If you were positioning for the dovish path discussed around the Jackson Hole pivot, this curve is the market telling you that path got repriced. The front end is where that argument gets settled, and the front end has moved the most.
2. The Arithmetic: Why the Damage Is Concentrated by Multiple
Now the part that actually determines portfolio impact. Every valuation on this blog reduces to the same identity — and the discount rate lives inside it:
Fair value = Earnings ÷ (r − g)
Where r is the discount rate and g the long-run growth rate. Invert it and the multiple you are willing to pay is simply 1 ÷ (r − g). A high multiple is not a mood — it is an arithmetic statement that r and g are close together. And when a denominator is small, adding to it hurts disproportionately.
Concretely: for a +100bp rise in r, the fair-value hit is
− (0.01 × P/E) ÷ (1 + 0.01 × P/E)
which produces this, and it is steeper than most people expect:
The Same Rate Move, Priced by Multiple
*Illustrative — shows the mechanism, not a forecast. Pure arithmetic from a growing-perpetuity model, holding earnings and growth fixed so the rate move is the only thing changing.
A 12× value stock gives up about 11% of fair value. A 30× compounder gives up 23%. A 40× growth name gives up 29%. Same rate move, roughly triple the damage — and none of it requires the company to miss a single earnings estimate.
This is why “rates up, stocks down” is too coarse to act on. The correct statement is rates up, high multiples down more, and the dispersion between them is usually the trade.
To make it concrete with something from this blog: the NVDA fair-value work I published used a 30.1× target multiple. Run this year’s 48bp move on the 10-year through that multiple and, mechanically, roughly 11% of fair value comes out — before a word is said about GPUs, competition or margins. That is the part of a drawdown that has nothing to do with the company.
3. Now Let Me Argue Against My Own Chart
That second chart is clean enough to be misleading, and I would rather point at the flaw than let a reader walk into it.
It holds g fixed, and g is not fixed. The arithmetic assumes only r moves. But rates rarely rise for no reason — they rise because growth or inflation surprised upward, which usually means nominal earnings are also being revised up. If g rises alongside r, the spread (r − g) barely widens and the multiple hit largely cancels. That is precisely why equities have historically tolerated rising rates during expansions and hated them during stagflationary scares: it is not the level of r that matters, it is whether g came with it.
And the front-end move argues both ways. A bear flattener driven by a stronger economy is the benign version — growth is showing up, so g is rising too. A bear flattener driven purely by a hawkish policy repricing against soft data is the ugly version: the discount rate goes up while the growth path goes down, and both terms of the fraction move against you at once. Which one is happening is a question about the incoming data, not about the curve, and I do not think the curve alone can settle it.
| What moves | What it usually means | Equity read |
|---|---|---|
| Bear flattener front end leads — this year |
Policy repriced tighter | Multiple compression concentrated in long-duration names; tolerable if earnings are also being revised up |
| Bear steepener long end leads |
Term premium, inflation or fiscal risk | The genuinely dangerous one — it raises the denominator for every asset at once |
| Bull steepener front end falls fastest |
Cuts priced in, often into weakness | Multiples get relief, but usually because earnings are about to be cut |
4. What Would Change My Mind
- The 30-year taking the lead. If the long end starts outrunning the front end — a steepener — my “this is a policy repricing” read is wrong and the more dangerous term-premium story is in play. That is the single number I would watch above all others.
- The 10-year breaking decisively above its past-year range. It sits at 4.67% against a 3.97%–4.75% band. A sustained break above that top is not just a level change, it is evidence the range regime itself has broken.
- Earnings revisions turning down while yields keep rising. That is the ugly-version confirmation: both terms of the fraction moving against equities together, which the second chart deliberately does not model.
- Dispersion failing to show up. If rates rise and high-multiple names don’t underperform low-multiple ones, then the discount-rate channel is not the dominant driver right now, and I should stop reasoning about the market through it.
Quick FAQ
Q. Should I rotate out of growth into value because rates are rising?
I can’t give you portfolio advice and I’m not licensed to. What the arithmetic supports is narrower and more useful: the relative sensitivity of high multiples to r is a mechanical fact, not a market opinion. Whether that is already priced in is a completely separate question, and the answer to it is not in this arithmetic.
Q. Is a 4.67% 10-year high by historical standards?
Not by long history — it is high relative to the 2010s, which is the period most current intuitions were formed in. That framing matters more than the level: a lot of what feels like “rates are extreme” is really “rates are unlike the decade I learned in.”
Q. Why did you use a perpetuity model instead of a full DCF?
Because the extra machinery would add precision without adding truth. A multi-stage DCF has more inputs I would have to assume, and the qualitative result — sensitivity scales with the multiple — is identical. When a simple model and a complex one agree on direction and magnitude, publishing the simple one is the honest choice.
💡 Quant Strategy & Takeaways
The 2-year has risen 73bp this year against 33bp on the 30-year: a policy repricing, not a term-premium panic. The damage lands by valuation — a 30× multiple loses roughly 23% of fair value per 100bp, a 12× multiple about 11%.
So watch the long end for the regime, and your own multiples for the magnitude. 🤖
Are you treating this move as a policy story or a term-premium story? I’d like to hear which, and why. 📈✨
Disclaimer: This article is quantitative research published for informational and educational purposes only. It is not financial advice or a recommendation to buy or sell any security. Yield data is as reported on the date stated; the valuation sensitivity shown is an illustrative model, not a forecast.
Comments
Post a Comment