Priced for Delivery

Why an earnings-led market is a different risk from an expensive one

With U.S. equities again trading near record highs, the instinct for investors is to focus on what could go wrong, and this month that list has a new entry at the top of it. The 10-year Treasury yield closed above 5% for the first time since 2007, having begun the year a little above 4%. A move of this magnitude in the risk-free rate raises the discount rate on every future cash flow and the return equities must offer to compete with bonds. It has also arrived alongside inflation that has stopped falling and a Federal Reserve that raised rates on 16 September 2026 for the first time since 2023. Set against a multiple that is high by the standards of the past three decades, and an earnings yield now only a quarter of a point above the 10-year, the case for a materially more defensive stance appears to write itself. The more useful question is whether the fundamental backdrop has deteriorated enough to justify acting on it. So far, the answer is not obvious.

The information and opinions contained herein have been compiled or arrived at from sources believed reliable, but no representation or warranty, express or implied, is made as to their accuracy or completeness. Analysis used data from Bloomberg and other sources. Neither the information nor any opinion expressed constitutes a solicitation. Any further disclosure, use, distribution, dissemination, or copying of this publication is prohibited. Performance data represents past performance and is not indicative of future performance.

An Earnings Rally, Taxed by Valuation

That earnings have led this market is now widely accepted. What gets less attention is how much of that growth the multiple has taken back. Over the twelve months to 15 September the S&P 500 returned 14.7% in price terms, while forward earnings rose 36.7% and the forward multiple fell 16.1%. Earnings growth alone would have lifted the index by more than a third; the fall in the multiple cut that to 14.7%. The multiple is still high, in the 71st percentile of its modern history, but the direction of travel has been down.

The pattern holds across other windows. Year to date the index is up 10.8% against forward earnings growth of 28.6% and a 13.8% de-rating. Stretch the window to two years and price is up 34.8% against earnings up 50.9%, with the multiple down 10.6% (see Exhibit 1). At every horizon we examine, the multiple has been a headwind. The forward P/E entered the year at 22.1x, peaked at 22.4x on 12 January, and now sits at 19.0x. Investors have been paying less for each dollar of earnings, even as those earnings grew. Consensus forecasts for the index as a whole have also risen every month for the past year, without a single monthly decline.

Recognising the rally as earnings-led changes what we should watch, and it changes what should worry us. If the market were being carried by the multiple, the relevant risk would be the discount rate, and the break through 5% described above would be close to decisive. October 2023 stopped at 4.99% without crossing it. Because the market is being carried by earnings, that rate move has already registered through the de-rating, and the live question is the durability of the earnings themselves.

Our caution is about the pace of earnings growth. Forward earnings growth of 36.7% year on year sits in the 98th percentile of readings since 1994. There have been only two prior episodes at or above 30%, in 2010 and from April 2021 to January 2022, which is too few to build a base rate on. What those episodes do suggest is that growth of this order does not collapse. Twelve months after such readings, earnings growth has run between 13% and 17% on the mean and median, comfortably ahead of the long-run average of 8%. The risk is not that earnings fail. It is that they decelerate from extraordinary to merely strong, and that a market accustomed to 37% treats 15% as a disappointment.

The de-rating tells us how much of that scepticism is already in the price. A compression of 16% over twelve months, 3.7 points off the multiple, against earnings growing at 36.7%, is consistent with a market that does not capitalise the current run-rate as permanent. Other readings of the same de-rating are available, and a higher discount rate and a higher risk premium would leave the same footprint, so we would not claim to have isolated the cause. Whatever the cause, consensus expects a further 21.7% on top of trailing earnings, and the index has declined to pay for it.

The Cushion, and what 1997 to 2000 actually teaches

What the de-rating has not done is restore a cushion. At 19.0 times the forward earnings yield is 5.26%, against a 10-year at 5.00%, leaving 26bp between the two (see Exhibit 2). Only 3% of trading days in the last twenty years, and 6% in the last ten, have offered stocks a margin this thin or thinner over bonds. Outside the 1993 to 2002 window, the spread has been this thin on only 156 trading days, every one of them in 2024, 2025 or 2026.

The late 1990s are the obvious comparison, and they cut both ways. Through 1997 to 2000 the forward earnings yield sat below the 10-year on 96.5% of trading days, averaging minus 1.18 percentage points against today's plus 26bp (at time of writing), and the index still rose 106% from the end of 1996 to its March 2000 high. A thin cushion is a poor timing signal, and we would not treat today's 26bp as one. How that episode ended is the less comfortable part. The index peaked on 24 March 2000 and fell 49.1% over the following two and a half years (dot-com bust). The spread had been inverted for years by then, so it was not the trigger, and earnings never collapsed. At the peak forward earnings were still growing 14.6% year on year. Twelve months later the forward estimate sat 1.4% below its peak level and the index was down 25.4%. Earnings simply stopped compounding, and that was sufficient, because a multiple at 25.3x stood more than eight points above its long-run average and had a great deal to give up. It gave up 6.2 points in the first year and 11.3 by the October 2002 trough, finishing at 14.0x.

Set against that episode, today compares well on two of three counts. At 19.0x the multiple sits 1.9 points above its long-run average, where the 2000 peak stood more than eight points above it, so there is far less air to come out. The cushion is positive rather than inverted, so equity holders are being paid something over Treasuries rather than accepting less. Against that, the earnings line has carried the entire return. In effect the market has swapped valuation support for earnings delivery, which leaves earnings less room to disappoint than when valuation offered a buffer.

Whose Earnings

The obvious objection is that index earnings are not one thing. The seven companies usually grouped as the Magnificent Seven account for 30.0% of consensus forward earnings for the index, against 28.0% at the end of last year, 25.0% at the end of 2024, 18.0% at the end of 2021 and 10.2% at the end of 2015. Concentration on this measure has nearly tripled in eleven years and it is still rising, because those seven grew forward earnings 51.3% over the past twelve months against 38.2% for the index as a whole. Nearly two thirds of the improvement nonetheless came from the other 493 companies, which is more broadly based than the concentration narrative usually allows.

The seven are not behaving as a group, however (see Exhibit 3). Nvidia alone produced 18.4% of the entire increase in index forward earnings over twelve months, half of the seven's combined contribution. Add Alphabet and Amazon and three companies account for 30.7%. The remaining four contributed about six per cent between them, and Tesla's contribution is negative. The revision paths tell the same story. Over twelve months consensus forward earnings for Nvidia rose 130%, while Tesla's fell 8%, and over the past three months Meta and Tesla have both been cut. Measured by earnings, Nvidia is carrying the group, Alphabet and Amazon are contributing strongly, and the other four are adding comparatively little.

What makes this less uncomfortable is that investors are not paying up for it. The seven trade at 22.3x forward against 19.4x for the index, a relative multiple of 1.145x. That is the 16th percentile of readings since 2015, against a peak of 1.607x. A thirty per cent share of index earnings carries a fifteen per cent premium on the multiple, and that premium sits near the bottom of its own range rather than the top (see Exhibit 4). The risk is not simply concentration itself, but the growing dependence of index earnings growth on a small number of companies. With the premium already modest, the threat from these names runs through earnings, and close to a fifth of the index's forward earnings growth now rests on the revision path of a single company.

For portfolios that argue for favouring companies where earnings estimates are still being revised higher, while limiting exposure to sectors whose valuations are particularly dependent on lower interest rates. Earnings revisions, more than the level of the 10-year Treasury yield itself, are what we will be watching.

PMI