Good economic news came out, yet yields jumped to a 20-year high and small caps fell. With only the Nasdaq near record highs, will this market carry on into next week?
2026-W39
A stretch where good economic data pushes yields higher
Wednesday's September composite PMI of 58.4 was the highest since July 2021, and input costs in the same survey rose at the steepest pace in four years (TechTimes, 2026-09-23). The market read the number not as a sign of economic strength but as grounds for the Fed to hike further. The 10-year yield jumped 13.89bp in a single day, the biggest daily rise since April 2025, and the probability of an October hike per CME FedWatch rose from 55% to 66.4% in one day (Kiplinger, 2026-09-23). One tally shows close to 95% pricing of a further hike by December (Discovery Alert, 2026-09).
The key is where the rise in yields came from. When yields rise on fears of an economic slowdown, all stocks fall together. This week, growth and costs pushed yields up together. So only stocks able to overcome a higher discount rate with earnings growth survived. AI chips met that condition; the rest did not.
Stocks are no longer cheaper than bonds
The equity risk premium (ERP) compares the expected return on stocks with bond yields. It is the S&P 500 earnings yield (the inverse of the P/E) minus the 10-year yield. In April this year, the earnings yield was about 4.6% and the 10-year 4.29%, narrowing the ERP to 30-35bp (ECM Source, 2026-04). Since then the 10-year has climbed nearly 90bp further to 5.18%. Assuming a forward P/E of 21-23x, the earnings yield is 4.3-4.8%, putting the ERP in negative territory of roughly -0.4 to -0.8 percentage points (calculated using the forward P/E ranges from ECM Source and RealInvestmentAdvice). After the ERP was last negative in 1999-2000, the S&P 500 fell nearly 50% over two years (ECM Source, 2026).
A negative ERP, however, does not tell you when the market will turn. It tells you how deep the decline can go when it does. As long as bonds pay a risk-free 5%, stocks have little room for excuses if earnings fall even slightly short of expectations.
Why small caps crack first
The gap between the Russell 2000 at -0.80% and the Nasdaq at +2.06% reflects a difference in debt structure. About 40% of Russell 2000 company debt is floating rate, versus less than 10% for S&P 500 companies. Some 40-46% of Russell 2000 companies cannot cover interest from operating income and survive by refinancing, and small-cap debt maturing this year totals $368 billion (BingX Research, 2026). Cash-rich megacap tech companies, by contrast, earn more interest income when rates rise. The same rise in rates is a cost for one side and income for the other.
The closest precedent: October 2023
The last time the 10-year hit 5% was October 2023 (5.021%). That month the S&P 500 slid about 10% from its July peak into correction territory (S&P Dow Jones Indices, 2023-10). The rebound came after yields turned. When the November Fed meeting signaled the end of hikes, the 10-year came down quickly and stocks rallied strongly through year-end.
This time the order is reversed. In October 2023 the Fed was wrapping up its hikes; now it has just resumed hiking on Sept. 16. That means the Fed is unlikely to supply the catalyst that brings yields down. If a pullback in yields comes, the trigger will be an external shock such as softer economic data or a shutdown.
Paths for next week
| Path | Conditions | 10-year | Stock market |
|---|---|---|---|
| Yields rise further | Data stays strong; Fed officials signal more hikes | Breaks intraweek high of 5.23% | Breadth narrows further. Weakness extends in the Russell 2000, utilities and REITs. The Nasdaq is also exposed to a spike in tech volatility |
| Range-bound | Micron earnings solid; shutdown patched up short term | 5.0-5.2% | This week's pattern holds. AI megacaps lead the index while the rest go nowhere |
| Yields pull back | Prolonged shutdown creates a data vacuum; hiring slows | Below 5.0% | Technical rebound in this week's laggard sectors. Small caps react the most |
The first path is the one that would break this week's pattern. The 6-point gap between the VIX at 14.87 and the Nasdaq volatility index VXN at 20.87 means the market as a whole is relaxed, but demand for insurance against a tech selloff is growing. Anxiety about the one pillar holding up the index is the first thing showing up in prices.
What investors should do
- Watch three numbers: whether the 10-year closes above 5.23%, whether VXN tops its intraweek high of 21.88, and whether the Russell 2000 breaks its weekly low of 2,810.75. If two or more happen at the same time, treat it as the first path.
- Keep the core indexes (SPY, QQQ), whose stop-loss cushions are ample in the 29% range. But don't add new money. When the expected return on stocks is below bond yields, chasing gains offers little reward.
- Hold off on small caps until the 10-year falls below 5.0%. A structure heavy in floating-rate debt won't change until rates come down.
- The semiconductor weighting has grown even within the core indexes. Before Micron's Sept. 30 earnings, calculate what share of your total equities is tech exposure across SMH, QQQ and XLK combined. All three react in the same direction to the same event.