Philip Coggan: The mean truth about alluring markets

Albert Einstein supposedly said that compound interest “is the greatest mathematical discovery of all time”. And it does play an enormously important part in investment returns. But reversion to the mean must run a close second.

Reversion to the mean does not apply for every individual stock or to every country. Some companies go bust; some countries descend into revolution; and investors lose all their capital. It is not, therefore, the case that what goes down must always come up. But it does seem to be the case that what goes up will eventually come down.

Individual companies cannot increase their profits or sales at 20 per cent forever. The corporate
sector as a whole cannot increase its profits faster than GDP indefinitely. Shares will not be continuously re-rated; in other words, share prices cannot keep outpacing profits or dividends.

Unfortunately, investors have a terrible tendency to extrapolate. They tend to assume that if an asset has recently been performing well, it will continue to do so. In its latest equity-gilt study, Barclays Capital showed there was a remarkable correlation between US equity mutual fund sales and stock market returns over the previous six months. The same was true of bond funds.

It is an odd process which suggests investors are rather confused about asset prices. It was very easy to get UK investors to buy equity individual savings accounts in early 2000, when the FTSE 100 index was well over 6,000; very difficult to get them to buy the same funds in early 2003, when the index was a little over 3,000.

It is as if shoppers decided to boycott the department store retail sales and wait until prices were higher in the spring.

Obviously, there is a psychological factor at work. Investors need to be persuaded to give up the safety of cash. They want to see share or bond prices rising before they take the plunge.

But this leads to very bad timing on their part. Barclays Capital finds that, after around 11 to 12 months, equity and bond returns tend to be negatively correlated. In other words, if shares and bonds have done well over the previous year, they are unlikely to produce positive returns over the next few months.

And this relationship is borne out by simple logic. Yield is an important component of investors’ return; and as prices rise, yields fall. So, other things being equal, a given price rise should lead to a fall in expected returns.

Barclays Capital has run the UK bond market data over the past 105 years and found that nominal yields of less than 4.3 per cent are not associated with positive future real (after inflation) returns. This is not encouraging, given that 10-year yields are around 4.2 per cent and long-dated yields were 3.9 per cent.

What about equities? Barclays found that returns over the previous 15 years were a very good contrary indicator of returns over the following 15 years. Accordingly, in 2000, after a bull market that had started back in 1982, investors should have been cautious about future returns. In fact, they were more confident than ever.

The negative correlation becomes all the greater when past returns have been either very high or very low. The reason is simple; long periods of low returns lead to low valuations (and thus higher expected returns) while strong markets lead to higher valuations, and lower expected returns.

Indeed, investors can make a good guess at future equity returns if they look at the price-earnings ratio. When that p/e is over 19, then future returns are likely to be very low; when the p/e is under 12, returns are likely to be high.

Tim Bond of Barclays reckons that since the current US p/e is 19, returns should be OK going forward. However, a note of caution is needed: he calculates that p/e on the basis of operating earnings, which can be skewed upwards.

If one uses a different measure, such as the cyclically-adjusted price-earnings ratio (based on an average of the last 10 years’ earnings), the p/e is 30, according to James Montier of Dresdner Kleinwort Wasserstein. That makes the market look as expensive as it was in the late 1920s and does not suggest returns from here will be good.

All this theorising is based on reversion to the mean. This does not imply that share prices return to the same level; there is a long-term upward trend, associated with economic growth. It does suggest that valuations return to the mean (although Jeremy Grantham of the fund management group GMO, by no means a bull, thinks there is a gentle upward trend in valuations, thanks to more stable economies).

And what is true at the market level is just as true of individual stocks. All stocks, however alluring, will eventually trade at the market average p/e of 15-20. So if you buy a share on a p/e of 100, you are fighting against a potential de-rating of 80-85 per cent. The company needs to be able to deliver an awful lot of growth to offset that; its earnings need to grow fivefold to sixfold.

In 2000, Sage, a perfectly good software company, traded on a p/e of more than 100. It is still a perfectly good company, but it now trades on a p/e of 25. Its shares, once 930p are now 280p.

So the lesson is: do not slant your portfolio towards those markets that have recently done well. Indeed, the mere act of avoiding them may save your portfolio from disaster; think Japan in the late 1980s or technology in early 2000.

philip.coggan@ft.com