The first question you might have, upon seeing that headline, is what’s the difference? Data is simply facts, without meaning or context. Information can be used as a basis to make decisions; it can be acted on.
A good example is an old radio comedy routine that went something like this: “And now for the baseball scores: 4–2, 5–1, 3–1, and in a blowout surprise, 9–0.” Without the team names attached to the scores, the data isn’t particularly useful to fans. Add the context—the names of the teams—and it becomes information, usable to settle bets, update team records, or make fun of your friends.
Putting P/E ratios in context
An analogy in the financial markets is price-earnings ratios. 17, 21, 6, 15—without knowing anything more, these are meaningless numbers. What is the growth rate of earnings? What is the typical ratio in the industry? What time period do those earnings cover? Is it unusual?
Consider the different types of P/E ratios mentioned in the press, which are often confused with each other:
Each has its strengths and weaknesses, as well as known biases that are built into the calculation. Looking at any one ratio can result in a bad decision, while comparing two different types may leave you scratching your head. (For more on this general topic, see my recent post about different ways to look at data.)
What can each type of P/E ratio tell us?
Analyses that use the forward P/E are a pet peeve of mine. Forward earnings estimates have a systemic upward bias; they are almost always high. Forward P/Es, therefore, make stocks look cheaper than they really are. (This may be why they’re often quoted by Wall Street.)
The trailing 12-month P/E is better, in that it reflects what has actually happened. But its weakness is assuming that the past year was typical. Most years are not typical, and values look much higher in bad years and lower in good years. Right now, for instance—with record-high profit margins, low interest rates, and high stock buybacks—earnings are higher than normal, and stocks look cheaper than they would in a more typical year.
The Shiller P/E tries to address the normality concern by using an average of earnings over 10 years, on the theory that good years and bad years will average out over time. As a result, it is a much smoother indicator of value, avoiding the big swings of the yearly figures, but it has the drawback of using out-of-date information. Another weakness is that it has been at historically high levels for more than the past 20 years, raising the question of whether current figures are really useful or whether the market has changed.
The Hussman P/E attempts to sidestep the issues associated with annual figures, estimates, and longer time periods by using the best annual earnings the company has ever achieved. Because of this, there are no comparison issues; you’re simply valuing the company based on the best it has done. This provides a stable and defensible E for the analysis, but it raises an obvious question: will the company ever get those earnings again, or improve on them?
Which P/E ratio is best?
Personally, I find the trailing 12-month P/E (for its use of the most recent actual data) and the Shiller P/E (for its inclusion of a full business cycle of earnings) to be useful. Both, however, present the difficulty of comparing current figures to historical data.
The information contained in these ratios can be used in portfolio planning to estimate future returns. At this point, given current valuations and historical patterns, those estimates are likely to be disappointing for most people (i.e., in the low- to mid-single digits). You can certainly conclude that the indicators are wrong, but viewing them as information and not just data, you should be prepared to demonstrate why you think that’s the case.