New research by Robin Greenwood, Andrei Shleifer and Yang You at Harvard University has established this. They looked at all the periods since 1928 in which any US stock market sector has more than doubled over two years. They found that "a sharp price increase of an industry portfolio does not, on average, predict unusually low returns going forward". "Many industries that have gone up in price a lot just keep going up", they say.
Much the same is true in the UK. I looked at all the instances between January 1988 and March 2016 in which a major FTSE sector has risen more than 50 per cent over a 12-month period: where there were sequences of such rises, I took the first. Across 25 sectors, there were 88 such cases. This allows us to answer the question: should a 50 per cent gain be a trigger to sell?
The answer's no. Across these 88 cases, the average price rise in the following 12 months was 9.8 per cent, which is a little above the average price gain for all periods of 8.1 per cent. However, falls were quite common: they happened in 34 of these 88 cases, which is almost 40 per cent.
Take, for example, telecoms. These rose 59 per cent in the 12 months to December 1999 and fell 34 per cent in the following 12 months. This would suggest that a big price rise is a sell signal. But telecoms also rose more than 50 per cent in the 12 months to February 1998, and gained another 56 per cent in the next 12 months. The latter instance is slightly more typical than the former.
A big price rise on its own is, therefore, no proof that there's a bubble. It can instead be a sign that prices have recovered from low points: several of my 88 cases came in the 2009-10 rebound from the financial crisis. Or it might instead be a sign that a sector has enjoyed great news to which investors have underreacted; all this evidence is consistent with the fact that there is momentum in share prices.
Right now, mining and engineering stocks have risen more than 50 per cent in the past 12 months. My evidence suggests that there's no compelling reason to sell them on this account - although the evidence is only around 60-40 in their favour.
None of this, however, is to say that bubbles don't exist. Experimental research tells us that egregious overpricing does happen. It's just that we need more evidence than price rises alone for this. Such evidence, say the Harvard economists, includes: increasing amounts of new share issues;, higher volatility, and especially big price rises for newer companies.
We also know that there's a useful rule to tell us when to ride bubbles and when to jump off. It's the 10-month (or 200-day) moving average rule proposed by Meb Faber of Cambria Investment Management. This tells us to sell when prices dip below that average. Doing this won't get you out of the bubble at the very peak: only luck will do that. But it will limit your losses as the bubble deflates and so produce good average profits over bubbles and bursts if applied rigorously. We can, therefore, do better than simply selling after big gains.
