has AI abolished gravity for share prices?

Here is a good phrase for suggesting a stock price has risen unsustainably: “Up like a rocket, down like a stick.”

It works nicely because gravity is an analogy commonly used in investment. Hence, all that “soaring”, “climbing”, “falling” and “plunging” which market reporters attribute to shares.

But “gravity” is just too widely understood to sound technocratic. So financial natives talk about “mean reversion” instead.

The aim of this article is to unpack and test the thinking behind mean reversion. The phenomenon may — or may not — humble the US’s high-flying, AI-related stocks. How should private investors view it?

In an everyday context, mean reversion would involve spitefully backing your car over your neighbour’s cat. In a financial setting, it implies that prices rise and fall around a true value to which they always return.

If you uniquely knew what this was, you could buy shares below true value and sell them when they rose above it. Mean reversion would vindicate both actions.

“But it is impossible to predict when mean reversion will occur,” says Joe Wiggins, a behavioural finance expert and investment research director at St James’s Place, a wealth manager.

Guillaume Rambourg, a private investor and philanthropist, adds: “Everyone who has tried to call mean reversion over the past five years has got it wrong.”

Financial pundits have nevertheless been invoking mean reversion because the S&P 500, the mother of all stock indices, has risen by 94 per cent over five years to all-time highs. US stocks now account for a quarter of the value of the world’s equities.

Stale though it is with repetition, the main reason is the barnstorming performance of US tech stocks, in particular the so-called Magnificent Seven, which includes Microsoft and Apple. These were buoyed first by the pandemic and latterly by enthusiasm for emerging artificial intelligence technology.

Hence recurring predictions of a “mean reversion event”. It is clear what market commentators are pointing to here: a fall in US stock prices.

But, to paraphrase Humphrey Bogart, nominal stock prices do not alone amount to a hill of beans in the crazy world of valuations. What matters to investors is the relationship between those prices and potential earnings. Indeed, commentary predicting market drops generally focuses on this relationship. Their favoured metric is Shiller’s Cape.

This is not a rainproof garment you wear on your bicycle after the police impound your car during an investigation into the death of your neighbour’s cat. Instead, it is the cyclically-adjusted price to earnings (Cape) ratio devised by Professor Robert Shiller.

It tells you how many times stock prices can be divided by average earnings per share, with some adjustment for economic ups and downs.

The number for the S&P 500 is currently around 38 times, not far from a pandemic-era peak of 44 times. Over the past 20 years, the average was 27 times. Perhaps that is the true value that Shiller’s Cape is supposed to revert to.

Market doomsayers tend to envisage mean reversion in price/earnings via a drop in stock prices because this would be the doomiest way for it to occur.

But rising earnings could also fulfil the prophecy more positively, via higher earnings. That, you could argue, is what the market is predicting, assuming it embodies anything more sophisticated than animal greed and fear.

This thought sent me scuttling off to check S&P Visible Alpha, a data service that collates a plethora of estimates by brokerage analysts. And voila! The consensus among stock-specific predictionists is that earnings will rise sufficiently for the price/earnings ratios of the Magnificent Seven to drop to humdrum levels over the next five years. The only exception is eternal outlier Tesla.

The multiple for AI chipmaker Nvidia would, for example, be just 16 times at today’s stock price and at 2029 predicted earnings. That compares with a current valuation of 48 times on 2024 earnings.

To hit Visible Alpha estimates, the Magnificent Seven would need to raise earnings at quite a clip. But rates of increase would not need to be any higher than those achieved by the tech giants in the previous five years. If that happens, AI would have abolished gravity — or at least the doomy version of it foreseen by some market watchers.

So case closed? Even as the police probe the suspicious circumstances of Fluffy’s demise?

US indices are worth holding at current levels if you believe AI is a transformative and potentially lucrative technology. If you don’t, you should sell or avoid

Unfortunately not. I respect brokerage analysts for the invaluable insights they provide. However, weary experience teaches us that when uncertainty is high, the comfort provided by numeric projections is usually illusory. A lot will happen in tech, much of it unforeseen, over the next five years.

Moreover, the financial services industry has a weakness for reverse engineering. That means starting with numbers that suit your customers and generating calculations to justify them. High tech stock prices could be those crowd-pleasing numbers. Earnings estimates produced by brokerage analysts could be the post-hoc justifications.

Earnings-based valuations would simply be, in this case, interesting but unreliable artefacts of the following binary choice. US indices are worth holding at current levels if you believe AI is a transformative and potentially lucrative technology. If you don’t, you should sell or avoid.

To me, automated systems with potential to substitute seamlessly for humans appear transformative in a way that virtual reality, the internet of things, cryptos and the metaverse never could.

I assume the US market has at least one good crash in it along the way. Of course, I have no idea when it will happen. Some businesses investing billions in AI would lose their shirts. Upsets inevitably result from the proliferation of disruptive technology. The clue is in the name.

It would be wrong to characterise that crash as a “mean reversion event”. It implies that  constants exist in finance as they were once deemed to do in such physical phenomena as gravity. In finance, as in physics, truth turns out to be more mercurial.

Shiller’s Cape might average 27 times with some consistency. But underlying stock prices, earnings numbers and constituent companies are in continual flux. Everything changes, as philosopher Heraclitus and boy band Take That asserted.

As for the unfortunate business with next door’s cat, rest easy: Financial Thinking is not the kind of column that squeals on its pals. Your secret is safe with me. 

Jonathan Guthrie is a journalist, adviser and former head of the Lex column. jonathanbuchananguthrie@gmail.com