Want to know the future? Don’t trust the stockmarket

This is the efficient-market hypothesis, the formal version of which says that investors, in aggregate, perfectly and promptly incorporate new information into asset prices. Those who invoke it can often mean something even stronger: that markets therefore provide the best possible forecasts of fundamentals like corporate earnings. In other words, the price is always right, as it surely would be if it were economists cantering around and making all the decisions.

Right now this Platonic ideal feels especially remote. Retail traders clamour for meme stocks, whipping up prices just to give short-sellers a thrashing. Shares in GameStop, an ailing video-game seller selected for such favour in 2020, are still worth more than 20 times as much as they were then. They have done about as well as Nvidia’s, the biggest beneficiary so far of the artificial-intelligence revolution. Nvidia’s fellow tech giants are racing to issue new stock and sell it to the public—a sure sign that they reckon the bull market is nearing its peak. Yet investors are still happily piling in.

Those financial economists who do visit the stables have known for nearly half a century that markets are far more volatile than they would be if new information were all that moved them. Robert Shiller, who won the Nobel prize in 2013, showed this for bond yields in a paper published in 1979 and for stock prices in 1981. Over the hundred years or so of data he studied, stock prices fell several times by much more than could have been justified even by a Depression-scale downturn. This made it implausible that investors were pricing shares based only on sober forecasts of their dividends.

More recently Mr Shiller’s intellectual heirs have helped explain why—aside from people’s occasional tendency to bolt off and join a stampede. The most persuasive theory, held increasingly by both researchers and academically minded investors, is the “inelastic-markets hypothesis”, coined by Xavier Gabaix of Harvard and Ralph Koijen of the University of Chicago. Its crux is that share prices, rather than being set by the dividends (or earnings) investors expect, are buffeted significantly and lastingly by capital flows. Messrs Gabaix and Koijen estimate that someone who buys $1-worth of shares with fresh cash pushes up aggregate stockmarket value by $3-8.

To see why, picture three types of investor. One is a return-chaser, who buys more shares when they are on a tear and sells when prices are falling (think of retail traders or trend-following hedge funds). The second maintains fixed asset allocations: 60% to stocks and 40% to bonds, say (think of a pension scheme). The third is a value investor, only interested in buying shares at rock-bottom after a crash (think of Warren Buffett). Squint, and this stylised mix looks rather like the groups that dominate real-world markets. Importantly, any arbitrageurs—who efficient-market enthusiasts imagine might smooth out distortions and reprice assets according to fundamentals—are few, and tightly constrained.

Now imagine a retirement saver who wants to buy shares in a bull market. They cannot hit up the return-chaser, who wants more themselves. The fixed allocator can sell only if prices rise, since they must maintain their 60/40 split. The value investor, too, will sell only if shares get more expensive and hence, to them, less attractive. So the capital flow sends stock prices up, regardless of what anyone thinks about future earnings.

If the thoroughbreds manning trading desks get frustrated by economists touting efficient markets, they in turn may smirk at this explanation. It does, after all, sound quite like “prices rise to match supply to demand”, without saying why demand rose to begin with.

That absence might in fact be a strength. Ordinary savers who buy stocks each payday do not generally base their purchases on forecasts of stellar corporate earnings. Nor, necessarily, do institutional investors who are told to aim for a set return target and have few better shots. Their actions nevertheless push share prices up. Betting that this might predict the future seems like a good way to lose your shirt.