Cyclical stocks and normalized earnings

Some businesses earn steady profits whatever the economy does. Cyclical businesses do not: their sales and profits rise and fall with the economic cycle, commodity prices or capital spending. Valuing them with a single year's earnings is one of the classic ways to go wrong, because the ratio that looks cheapest often arrives just before profits fall.

What makes a company cyclical

A cyclical company sells something whose demand swings more than the economy as a whole. Typical examples include automakers, steel and chemical producers, airlines, homebuilders, semiconductor makers, oil and gas producers and miners. The common features are:

• Discretionary or deferrable demand. Customers can postpone a car, a house or a factory upgrade.
• Commodity-like pricing. Prices swing with supply and demand, and producers have little control.
• High fixed costs. Fixed costs magnify the effect of changing sales on profit, a property known as operating leverage.
• Capital intensity. Capacity takes years to build, so supply tends to overshoot demand.

Defensive companies, such as makers of everyday food and household goods, show much smaller swings. The Market Compass and the article on economic regimes describe how sectors have tended to behave in different phases of the cycle.

The earnings trap

The price-to-earnings ratio divides the price by one year of earnings. For a cyclical stock that single year can be at either extreme:

• At the peak, earnings are abnormally high, so the P/E looks low. The stock appears cheap just when profits are about to shrink.
• At the trough, earnings are depressed or negative, so the P/E looks very high or meaningless. The stock appears expensive just when profits are about to recover.

Markets are aware of this, which is why cyclical stocks often trade at low multiples near the top of the cycle and at high multiples near the bottom. Buying on a low P/E at the peak can turn out badly, a trap that is easy to fall into without a longer view.

Normalizing earnings

Normalizing means replacing the current year's profit with a level the business can plausibly earn over a full cycle. Common approaches:

1. Average earnings over a full cycle. Use the average of seven to ten years of earnings per share or operating profit, as long as that span covers both good and bad periods. The same idea underlies the cyclically adjusted P/E that is used for broad market indexes, which averages ten years of inflation-adjusted earnings.
2. Mid-cycle margin on current sales. Estimate the average operating margin across the cycle and multiply it by today's revenue. This adjusts for company growth over time.
3. Fit a trend. Run a line through past profits rather than relying on a single point.

Normalized profit can then be used in the usual tools, including earnings power value and enterprise value multiples such as EV/EBIT.

Example (illustrative numbers)

A steel producer reported these earnings per share over five years, and the stock price is $60:

On year 3 earnings the stock looks very cheap at 6 times. On year 5 earnings it looks very expensive at 60 times. The five-year average EPS is $24 divided by 5 = $4.80, so the normalized P/E is $60 / $4.80 = 12.5. That figure sits between the extremes and better reflects what the company earns across a full cycle. Whether 12.5 is a good price depends on how it compares with peers and on the company's returns, but at least the starting point is not distorted by the timing of the snapshot.

How to read it

Rules of thumb:

• Ask where in the cycle the latest earnings sit. Compare current margins and returns with the company's own multi-year range.
• For cyclical companies, a low P/E is a reason for caution, not comfort, unless earnings are clearly depressed.
• Cash-flow and balance-sheet measures are equally cyclical. Check debt in particular, because leverage hurts most when profits fall.
• Asset-based measures such as price to book can be more stable references for capital-heavy businesses, though they have their own limits.

Common mistakes and limits

• Assuming every decline is cyclical. Some downturns are structural, for example when a product becomes obsolete. Averaging past good years then overstates what the business can earn.
• Choosing the window to suit the answer. A seven-year window that starts at a trough or ends at a peak biases the average.
• Ignoring changes in the business. Acquisitions, divestitures and new capacity change the baseline.
• Forgetting inflation. Averaging old nominal profits understates today's scale.
• False precision. Normalized earnings are an estimate, and the result belongs in a range, not a point. See why fair value is a range.

Where to see it on StocksXray

StocksXray ranks each metric as a percentile within the company's own sector, so a cyclical company is compared with other cyclicals rather than with defensive businesses. That does not remove the peak and trough problem, though: the whole sector may be at its peak. Browse sectors such as energy and materials with that in mind.

Key takeaways

• Cyclical profits swing with the economy, so a single year's P/E can be misleading in both directions.
• Normalizing replaces the current year with a mid-cycle level, using multi-year averages or a mid-cycle margin on current sales.
• A low P/E at peak earnings is a classic trap, and a high P/E at trough earnings can hide an opportunity or a real problem.
• Normalization is judgment, so test several windows and treat the result as an estimate.