Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Good morning. Yesterday was a remarkable day for US stocks. Anthropic launched new AI tools for automating legal work, and in response software and analytics groups serving all sorts of white-collar industries took a beating. Software and professional services companies in the S&P 500 fell by over 3 per cent. Did tech giants like Alphabet and Microsoft, with their epic AI budgets, rise? On the contrary. The winners, instead, were old economy sectors such as energy, telecoms, staples and materials. Things are changing out there. Watch your step, and email us: unhedged@ft.com.
Market concentration is fine
The Magnificent 7 tech stocks no longer dominate the S&P 500, or investors’ attention, in quite the way they did a few months ago. But the index is still very narrow. The blue line on the chart below shows the cumulative contribution to the S&P 500’s total market capitalisation, starting at the left with the contribution of the single largest company (Nvidia, at 7 per cent of the total), then moving on to the two largest combined (Nvidia plus Alphabet, 14 per cent), then the three largest combined, and so on until all 500 are included. The pink line does the same for cumulative net income.
Just six companies contribute a third of the S&P 500’s market cap; just 62 make up two-thirds of it. For net income, the curve has a very similar shape, but is slightly less steep. Those six groups make up 27 per cent of the net income, and the 62 contribute 63 per cent. That is to say, the biggest companies tend to have somewhat higher valuations — price/income ratios — than the smaller ones.
The question that lots of pundits have been asking is whether the high concentration in market cap makes the market riskier. Should investors be underexposed to the largest companies or sectors, relative to market weights? A new paper called “Magnificent, but Not Extraordinary” by Per Bye, Jens Soerlie Kvaerner and Bas Werker argues that concentration is a normal condition in markets and does not indicate elevated risk.
The first part of the paper is built around a time series of all the companies traded on major US exchanges since 1926, including their market cap and various measures of profit. Here are the series for the market weight of the largest single company in the market, and of the seven largest combined:
The authors write:
We see that there is nothing special with the relative importance of the “Magnificent Seven”. For example, from the 1930s to the 1960s, seven firms held comparable shares. The peak occurred in May 1932, when AT&T, Standard Oil Company, Consolidated Gas Company of New York, General Motors, DuPont, R.J. Reynolds Tobacco Company, and United Gas Improvement Company together accounted for roughly one-third of total value.
The authors also point out that fundamentals such as revenue and earnings tend to track market cap concentration — but that the correlation is loose and shifts cyclically. The blue line in the chart below shows the share of all companies that make up a third of the total market value; today the share is at historical lows at less than half a per cent, as shown on the right axis. Meanwhile, the largest third of groups by market cap account for a historically low share of total public company revenue, profit and cash flow, at about a fifth, as shown on the left axis:
This is a slightly tricky chart, but it shows a simple relationship: As market cap concentration hits extremes (the blue line falls), the share of revenue, earnings and cash flow held by the largest companies tends to fall (the other lines fall). This is the same pattern we see in the chart we started with: economic fundamentals are concentrated in a small number of groups, but market capitalisation is even more concentrated — meaning that the valuations of the largest companies are high.
So high market cap concentration tends to be accompanied by high market valuations. And we know (as we recently discussed here) that high market valuations predict low long-term market returns. So does high concentration also predict low long-term returns? Not independently, Bye told me:
In isolation, market concentration is also negatively predictive of returns (high concentration, low subsequent returns), but not if you control for valuations . . . keeping valuations fixed, higher concentration is actually correlated with higher future returns! (We should be careful about statistical significance here because I’ve not corrected the standard errors for time-series correlation) . . .
It’s not that we argue we’re not in an AI bubble, but that one should be careful about drawing conclusions based on market concentration alone. Valuations . . . are likely more informative about the state of the market.
The second part of the paper reinforces the point about the normality of market concentration using a model based in “a standard geometric Brownian motion diffusion process” incorporating “a common market factor and firm-specific shocks”. I don’t know what that means, so I asked Bye. He dumbed it down for me as follows:
This is perhaps the most standard model of asset prices, and it matches the levels of concentration we observe in the data. We do not need an “exotic” set-up to match this empirical regularity. In this model, returns follow a random process. Think of companies’ market values as constantly being nudged in the form of productivity and innovation shocks, good and bad leadership, luck and misfortune. Over time, most companies remain small (their positive and negative shocks cancel out) while a few companies receive an abundance of positive shocks and grow large.
In short, if you set up a toy model of a stock market and just let it run, it too will tend towards concentration. The takeaway is simple: concentration, considered separately from high valuations, does not signal higher market risk.
One good read
FT Unhedged podcast
Can’t get enough of Unhedged? Listen to our new podcast, for a 15-minute dive into the latest markets news and financial headlines, twice a week. Catch up on past editions of the newsletter here.
