Lesson 2 of 9
Normalised earnings for a cyclical
A cyclical's earnings peak and trough with its industry. Value it on a mid-cycle figure, never on the year you happen to be reading.
By Javier Sanz
Founder of ValueScreener. Built Ninety Nine, a securities services provider, and sold it to a tier-one global bank; ran operations and internationalization at Alpaca, the brokerage API.
Who writes thisEditorial processLinkedIn (opens in a new tab)X (opens in a new tab)
In one minute
- A cyclical's earnings swing with its industry, so one year's profit is rarely a fair figure to value it on.
- Normalising replaces the reported figure with a mid-cycle one: an average margin, or a multi-year average, applied to today's business.
- Valuing a cyclical at its peak margin looks cheap and valuing it at its trough looks expensive, but neither read is right.
A cyclical company's profit rises and falls with an industry cycle it does not control. Valuing it on the latest reported year prices the business at whatever point of the cycle the filing happened to land on, which is a fact about timing, not about the business.
Reported profit against a normalised line
- Reported profit
- Normalised profit
Show the numbers
| Reported profit | Normalised profit | |
|---|---|---|
| Year 1 | $170m | $120m |
| Year 2 | $150m | $120m |
| Year 3 | $90m | $120m |
| Year 4 | $40m | $120m |
| Year 5 | $60m | $120m |
| Year 6 | $100m | $120m |
| Year 7 | $140m | $120m |
| Year 8 | $200m | $120m |
| Year 9 | $150m | $120m |
| Year 10 | $100m | $120m |
Why one year misleads
At the top of a cycle, margins are wide because demand outruns capacity. Earnings look large, the P/E looks low, and the company looks cheap on a figure that is about to fall. At the trough, the opposite: margins are thin, earnings look small, the P/E looks high, and the same company looks expensive on a figure that is about to recover. Graham made this point in chapter 12 of The Intelligent Investor: a single year's per-share earnings carry whatever the year happened to contain, which is why he preferred an average. His defensive test in chapter 14 capped the price at 15 times the average earnings of the past three years.
Two ways to normalise
The first method holds the business at its current size and asks what it earns through a full cycle at that size. Take operating margin for each of the last ten filed years, average it, and apply the average to the most recent year's revenue.
The second method averages the earnings themselves rather than the margin. This suits a business whose revenue is itself cyclical rather than steadily growing, where current revenue is a cycle position rather than a stable size to apply a margin to. Neither is a default. A company that has grown steadily fits the first; one whose revenue swings with the cycle fits the second.
Ten years, not one
Shiller's cyclically adjusted price to earnings ratio, developed with John Campbell, takes the same idea further: ten years of inflation-adjusted earnings, averaged, so a company does not look cheapest exactly when its margin peaks. Marks makes the underlying habit explicit in Mastering the Market Cycle: know where the cycle stands before you read a single year's figure as if it were the steady state.
The margin, not the multiple, does the work
Normalising changes the E in a P/E, not the P. A ten times multiple on a trough-year profit and a ten times multiple on a normalised profit are not the same statement about the price, because the denominators are different numbers. State which earnings a multiple divides by, or the multiple means nothing.
Check yourself
A cyclical company's margin has run 4, 12 and 20 percent at the trough, average and peak of its cycle. Which margin should value the business at its current size?
- 4 percent, to be conservative
- 12 percent, the cycle average
- 20 percent, the most recent year
Show the answer
Answer: B.
The cycle average removes the timing of the filing from the figure. The trough and the peak are both real, but neither is the steady state.Why did Graham average three years of earnings in his defensive tests rather than use the latest year alone?
- Three years is easier to find in a filing
- One year carries whatever that year happened to contain
- Averaging always raises the earnings figure
Show the answer
Answer: B.
A single year can sit anywhere in a cycle or carry a one-off item. Averaging is Graham's answer to not knowing where in the cycle a given year fell.A company's revenue is itself highly cyclical, doubling and halving with demand. Which normalising method fits better?
- Average margin applied to current revenue
- Average earnings across the cycle
- The latest reported year, unchanged
Show the answer
Answer: B.
When revenue itself is a cycle position rather than a stable size, averaging the earnings directly avoids applying a margin to a revenue figure that will not hold.Sources
- Graham, The Intelligent Investor, revised edition 1973, chapter 12, on per-share earnings and why a single year misleads.
- Shiller, Irrational Exuberance, 2000, and the cyclically adjusted price to earnings ratio developed with John Campbell.
- Marks, Mastering the Market Cycle, 2018, on knowing where a cycle stands before reading a single year's figure.
What this leaves out
- The worked example uses round invented margins so the arithmetic can be checked by hand.
- Ten filed years is the most fundamental history this product holds. An industry whose cycle runs longer than that cannot be fully normalised from the history on file here.
- Normalising chooses which years represent the business. That choice is a judgement, and two readers can normalise the same company differently.
- Nothing here is advice about any security.