The price-to-earnings ratio

The price-to-earnings ratio is the share price divided by earnings per share. A company trading at $50 with earnings of $2.50 a share has a P/E of 20, usually written 20x, meaning the price is twenty times one year of earnings.

Inverted, the same figure gives the earnings yield: 20x is 5%. That form is the more useful one when comparing a share against a bond, because both are then expressed as a percentage return.

Three versions of the ratio are in common use, and they differ only in which earnings go into the denominator. Mixing them is the most frequent error made with the P/E, because all three are quoted as a bare multiple and nothing on the number itself says which one it is.

A trailing P/E divides by the last twelve months of reported earnings. A forward P/E divides by an analyst estimate of the next twelve months, so it is a forecast and it is usually the lower of the two, because estimates generally assume growth. The cyclically adjusted ratio, published by Robert Shiller and usually written CAPE, divides by average inflation-adjusted earnings over ten years, which removes the effect of where a company or an index sits in its cycle.

Figures on company pages here are trailing twelve month unless labelled otherwise.

A company with no profit has no meaningful P/E. The arithmetic returns a negative number, which records an absence of earnings rather than a low valuation, and most screeners exclude negatives for that reason.

Earnings are an accounting figure and contain one-off items: a legal settlement, an asset sale, a writedown. A company earning $2.00 a share that books a $1.00 one-off gain reports $3.00, and its P/E at a $50 price falls from 25x to 17x without the business having changed. The following year it rises back, and a chart of the ratio shows a dip that never happened operationally.

Multiples are also not comparable across industries. Banks, insurers, utilities, software companies and miners trade at structurally different levels because their earnings differ in durability and in how much capital they consume to grow. A 12x utility is not cheaper than a 30x software company.

The ratio breaks down entirely at the bottom of a cycle in a cyclical industry, where it does the opposite of what a reader expects. Earnings collapse faster than the price, so the P/E is at its highest exactly when the shares are at their cheapest, and it is at its lowest at the peak of the cycle when earnings are unrepeatable. Miners, homebuilders and semiconductor companies are the usual examples.

There is no threshold that separates cheap from expensive. The S&P 500 has averaged somewhere in the mid to high teens over the long run, with wide variation by period, and a sector average is a more relevant comparison than an index one.

A high multiple reflects an expectation that earnings will grow. A low one reflects the absence of that expectation, or doubt that current earnings will persist. Low multiples cluster in industries in decline, so the cheapest ratios on a screen are usually cheap for reasons the ratio itself does not show.

Two comparisons carry most of the information. The first is against the company's own history: a business that has traded between 15x and 20x for a decade and now trades at 28x is being priced for something different from before, and the question is what. The second is against direct competitors on the same basis, trailing against trailing, in the same currency and the same accounting standard.

A P/E on its own says nothing about growth, which is the main thing it is being used to judge. The PEG ratio divides the P/E by the expected annual earnings growth rate expressed as a whole number, so a company at 30x growing 20% a year has a PEG of 1.5.

The convention, which comes from Peter Lynch, treats a PEG near 1.0 as a multiple in line with growth. It is a rule of thumb rather than a valuation method, and it inherits two problems. The growth rate is a forecast, usually the same analyst estimates that produce the forward P/E, and the ratio has no way to distinguish growth that is cheap to produce from growth bought with heavy capital spending or acquisitions.

It is most useful as a sanity check in the other direction: if a multiple only makes sense at a growth rate the company has never achieved, the multiple is the assumption being made.

Comparing a trailing P/E against a forward one. The forward figure is almost always lower, so the comparison manufactures a difference that is entirely methodological.

Reading a low P/E as cheap without asking what the earnings are. A company on 6x whose earnings are about to halve is on 12x, and the market usually knows before the ratio does.

Comparing across industries, or across countries with different accounting treatment and different tax rates.

Setting a per-share figure against a price on a different basis. Where a company has more than one share class, reported EPS may be per share of a class that is not the one quoted, which produces a P/E wrong by the conversion factor rather than a cheap valuation.

Treating a single quarter as a change in the multiple. Trailing earnings move a quarter at a time as old quarters drop out, so the ratio moves even when neither the price nor the business does.