p/e => e should be average over previous 10 years
==
The Long View: How p/e multiples can lead the sheep astray
By John Authers
Published: March 2 2007 19:22 | Last updated: March 2 2007 19:22
How expensive is a stock? It is a deceptively simple question that often leads to complicated answers. In turmoil such as that which struck world markets this week, it can seem close to irrelevant. When a wave of risk aversion suddenly breaks, a stock is worth as much as someone else will pay for it in the heat of the moment. Mature consideration of underlying value is left to one side.
But beyond the shortest term, valuation matters. If it becomes possible to buy stocks for less than they are worth, the chances that share prices will begin to recover is that much greater. If you are investing for the long term, you should invest even if prices take a while to return to some fair value.
The problem is working out how to tell if a stock is expensive. While there is no correct answer, one concept is unavoidable: the price/earnings ratio. When you buy a share, you buy a right to a share in the earnings that that company produces. The multiple you pay for those earnings must be a factor in deciding if a stock is expensive.
It has been recognised as such at least since Benjamin Graham, the founder of value investing, wrote Security Analysis with David Dodd in the early 1930s. The first of several factors he used to decide whether to buy a stock was its p/e multiple. It remains one of only two valuation measures in the Financial Times’ share price listings.
When p/e multiples are sending an emphatic signal about a stock, or about the market, it is seldom wise to ignore it. But there are reasons to fear that it could be misleading to follow an apparent strong “buying” signal sent out by the multiples on stocks, particularly those in the UK.
The most popular models using historic p/e ratios – measured by dividing the current share price by the most recent annual earnings – compare those multiples to historic averages, and to bond yields. When yields are high, and the relative risk-free return you can earn is relatively attractive, the argument is that p/es should be lower. When bond yields fall, p/es can rise.
Similar logic looks at historic averages. When p/e multiples are wildly above historic norms, stocks are probably too expensive.
Using either approach, stocks in the developed markets look cheap at the moment. In the case of the UK’s FTSE 100, which has a 23-year history, p/e ratios even reached a historic low of below 12 less than six months ago. The multiple on the S&P 500 reached 18, before dropping to 17 during this week’s turmoil. Its average over the past 20 years is 23.
During the long bullish run for stocks that was so suddenly interrupted this week, this was a vital factor in favour of the optimists. There was room for earnings multiples to expand. Even if earnings growth slowed, prices could carry on gaining thanks to multiple expansion. Relatively low multiples were at least a reassuring sign that the market was not too far ahead of itself. That logic, unfortunately, is flawed.
Earnings data from companies are “lumpy” and can vary widely from year to year, away from the trend. To be useful, a valuation system must take this into account. Also, it would be rational to expect p/es to follow the profit cycle to some extent. When times are good, and profits are high, multiples should be relatively low because analysts should know that profits will not continue at that level.
In the case of the FTSE 100, which looked so cheap late last year, the index is heavily weighted towards energy and resources groups that were coming off historically high profits which investors knew would be difficult to sustain.
The S&P 500’s earnings have completed their 14th successive quarter of growing by more than 10 per cent. P/es should be relatively low when the chances that earnings growth will slacken look relatively high. Peg (p/e over growth) ratios help capture this. A high p/e does not look so bad if it is divided by a high growth rate.
This ratio stayed stable for several years after the end of the tech bubble in 2000, but research by London’s Absolute Strategy Research showed that the Peg ratio had reached its highest level in five years, and suggested – correctly, it now appears – that this was a warning of a correction.
Graham’s work suggested that the “e” of the p/e should be the average over the previous 10 years. That begins to knock off the edges of lumpy earnings data, and include all points of the profit cycle.
Yale management school’s Robert Shiller showed in 1996 that the p/e of the index had predictive power over a point 10 years in the future, if the divisor was an average of real earnings over the past 30 years. This turned out to be a powerful predictive tool: when the p/e so measured was significantly above its long-term average, the chances were strong that it would correct back to the norm.
This has been dubbed the cyclically adjusted p/e. Andrew Smithers, the London-based analyst, used this model to turn on its head the orthodoxy that stocks look cheap. Far from being at historically cheap levels, as the one-year p/e on the FTSE 100 implied, he found that when p/es were cyclically adjusted, UK stocks were 68 per cent overvalued compared with their average since 1899.
This view is unusually bearish. But the market confusion of the past week seems to bear out that low p/es, on their own, are not much of a source of comfort.
john.authers@ft.com
Copyright The Financial Times Limited 2007
Comments