Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
This article is an on-site version of our Unhedged newsletter. Sign up here to get the newsletter sent straight to your inbox every weekday
Good morning. If forecasters are right, tomorrow will bring the first consumer price inflation report with a 2-handle headline number since 2021. The market may or may not care, but it’s good to remember how far inflation has come. Email us: [email protected] and [email protected].
We were wrong about the Mag 7’s valuation
Last week I (Rob) argued that if the Magnificent 7 tech stocks are overvalued, the S&P 500 is overvalued, too. I no longer think that is true, or at least it is way less true than I thought it was. I changed my mind because several readers pointed out to me I was making an analytical mistake in how I looked at the Mag 7’s valuation ratios, and once I corrected my error, the numbers look meaningfully different. Thanks in particular to Gerard Minack and Kristoph Gleich for pointing out the issue.
The problem is that I used weighted average price/earnings ratios for both the Mag 7 and rest of the index. On that method, the Mag 7 trades at about a 20 per cent valuation premium to the rest of the index. But this is distorted: you need to use a weighted harmonic mean, instead of the standard arithmetic average, to get the right answer. And using a harmonic mean, the premium is about 50 per cent. The chart below shows the difference:
Why? An average or arithmetic mean of P/E ratios naturally over-represents higher P/E ratios (if you are not interested in the arithmetic of why, you can skip the rest of this paragraph). Think of it this way: you own three stocks, each with a market cap of $100 (using market cap instead of price per share here just to keep things simple). They have expected earnings of $10, $10, and $1, so their price/earnings ratios are 10, 10, and 100. The arithmetic mean P/E of the portfolio is 40. But that’s not a representative number. Think of the portfolio as a single company, with a market cap of $300 and earnings of $21. The P/E of that company is 14 — which is the unweighted harmonic mean of 10, 10, and 100 (the formula is 3/(1/10+1/10+1/100)).
So the tale of the tape, using harmonic means weighted by market cap, is this:
The P/E ratio of the S&P 500, using earnings estimates for the next four quarters and excluding four companies with negative expected earnings, is 21.
The forward P/E of the Mag 7 is 31.
The rest of the S&P 500, excluding the Mag 7 and four stocks with negative expected earnings, has a forward P/E of 19.
Financial stocks, I would argue, are not properly measured with P/E ratios, and are a terrible comparison with tech stocks. So excluding financials and the Mag 7, the forward P/E of the rest of the index is 20.
Or in chart form:
So the simple valuation question is: is it worth paying 50 per cent more for the Mag 7, with their high growth, increasing returns, barriers to entry and momentum? Forced to answer in those simple terms, my guess would be no, it’s not. It could well be that the Mag 7 are overvalued and the rest of the index is close to fair value.
Looking at changes in valuation multiples over time, rather than current valuation levels, makes the Mag 7 look similarly expensive, as Roger Aliaga-Diaz and Ian Kresnak of Vanguard noted in an email. In a recent analysis, they started with the 43 per cent expansion in S&P 500 multiples since late 2018, and broke it into three categories: Mag 7, tech ex-Mag 7, and everything else. Of that 43 per cent expansion, two-thirds came from the Mag 7 and 83 per cent came from all of tech (including the seven). How sector-skewed multiple expansion has been strongly suggests Mag 7 valuations (and perhaps tech in general) have departed from the market.
There are subtleties and complications, though. Given the chance to disaggregate the Mag 7, I’d argue that Google and Meta don’t look terribly expensive relative to historical and expected growth; while Nvidia, Microsoft and Tesla do look very pricey. With Amazon and Apple, it’s very difficult to say: Apple has low expected growth for the price, but deserves a big premium for its defensive characteristics; Amazon’s valuation is hard to read because its underlying profit margin is hard to determine.
Meanwhile, you also have to ask if 20 seems expensive for the rest of the index. Absolute valuation is harder than relative valuation, though. One benchmark would be to compare the index’s earnings yield (the reciprocal of its P/E ratio) to available fixed income yields. One must say the S&P’s yield (5 per cent) does not look great when risk-free two-year Treasuries yield 4.5 per cent, the 30-year yield 4.4 per cent and it’s easy to pick up another point of yield with an investment grade corporate bond. This is the point of Robert Shiller’s “excess Cape yield” which is the cyclically adjusted S&P earnings yield less the 10-year real treasury yield. It is correlated with long term returns, and right now suggests that long term returns will be middling:
Someone with a lot of time on their hands should recalculate the excess Cape yield without the Mag 7 or even for the equal-weight S&P 500 index.
(Valuation, it is always worth noting, has no correlation with short term returns; current valuation implies nothing whatsoever about next year’s performance).
Summing up. The Mag 7 trades at a big premium to the rest of the index. Whether that premium is justified is a hard question to answer for those seven stocks as a group; each stock has its own story. The index as a whole, meanwhile, looks priced for middling-to-poor long-term returns. (Armstrong & Wu)