THE ECONOMIST: Lies, damned lies and stock markets — Why stock indices no longer reflect equity reality

Stock markets, Wall Street-wary politicians and like-minded commentators repeat ad nauseam, are not the real economy. The S&P 500 index of America’s biggest firms is up by 12 per cent this year, setting one all-time high after another.

Meanwhile, they point out, American GDP growth has slowed, employment has stalled and consumers feel more downbeat than at just about any point since the University of Michigan started keeping track in 1952.

All true — and nauseatingly familiar. Less remarked is the fact that the S&P 500 and the world’s other benchmark indices are no longer the real stock market, either.

Rather than mirroring the universe of domestic equities, they increasingly reflect the fortunes of a few corporate giants, which in turn mostly rise and fall with the unstable outlook for the artificial-intelligence revolution. This is making many benchmarks — and, by extension, the stock portfolios and pension pots of investors everywhere — considerably more volatile.

Take the S&P 500. The biggest constituent, Nvidia, accounts for 8 per cent of its value. Imagine the $US5.3 trillion ($7.48t chip-designer’s share price moves up or down by 3 per cent (as has happened 30 times this year) and the other 499 stocks are flat.

The index then shifts by 0.25 per cent in the same direction. No single name in the S&P 500’s history has packed a bigger punch.

Yet in comparison with the index it is a featherweight next to TSMC, which manufactures Nvidia’s AI processors. The semiconductor hulk, whose market capitalisation has doubled to $US2 trillion over the past year, accounts for over 40 per cent of the Taiwan Stock Exchange’s TAIEX index.

To let investors partake in TSMC’s success, last year the country’s markets regulator dropped the 20-year-old rule that index-tracking exchange-traded funds (ETFs) hold no more than 30 per cent of their assets in a single stock. As a result the TAIEX, which comprises just over 1000 stocks, is behaving like an equal-weighted index of just six firms (or fewer if you consider that other big Taiwanese companies like MediaTek and Foxconn are on the same AI steroids as TSMC and their movements are therefore partly correlated).

Things are only slightly less top-heavy in South Korea. The same bit of financial maths (which, for the curious, involves calculating the inverse of the Herfindahl-Hirschman index for component stocks) implies that the 830-strong KOSPI benchmark acts as if it contained just 11 firms besides the country’s two muscular memory-chip champions, Samsung Electronics and SK Hynix.

The pair’s combined weight in the KOSPI briefly exceeded 40 per cent in June, up from a quarter, give or take, in the preceding few years. (South Koreans’ love of leveraged ETFs, which use borrowed money to amplify gains but also losses, increases the KOSPI’s jumpiness.)

Between May and July, as traders constantly repriced the size of global AI spending, including on memory chips, the KOSPI’s volatility was four times last year’s average. The index moved by 5 per cent or more once every three sessions, on average. On July 28, it sank by 11 per cent.

Three days later it soared by 18 per cent. In the first six months of 2026 the Korea Exchange had to suspend trading five times and curb it on nearly 30 other occasions, more than in the whole of 2008, at the height of the global financial crisis.

Concentration and volatility need not go hand in hand. Nearly half the value of the broad Swiss index, SPI, sits in five firms: Roche, Novartis, Nestlé, ABB and UBS. But their industries are varied, and representative of Switzerland’s listed firms.

Almost one in two Swiss public companies makes drugs, food or drink, machinery or loans. In Taiwan and South Korea, by contrast, only one in six and one in 200, respectively, churn out chips or other technology hardware.

The top three Swiss stocks are also stodgy ones with stable earnings and regular dividends — and so inherently less skittish than tech shares. AI will transform them, as it will all sectors. But their fortunes are not hitched to the vagaries of the AI capex cycle.

Like a championship belt, dominance can be forfeited. Novo Nordisk took over the Danish bourse thanks to Ozempic, then shed its index weight as Eli Lilly’s better obesity drug ate its lunch.

That serves as a reminder that the stock market aims to capture investors’ best collective guess at tomorrow’s equity reality. It mirrors today’s only when the future is likely to look much like the present. In the age of AI, and of abundant private capital that feeds a long roster of unlisted pretenders which benchmarks omit, that is increasingly not the case.