P/E Ratio for Cyclical Stocks

The P/E ratio (stock price divided by earnings per share) is dangerously misleading for cyclical stocks. When earnings are artificially inflated at cycle peaks, the P/E looks cheap but the stock is overpriced; when earnings crater at troughs, the P/E looks expensive but the stock may be cheap. Analysts fix this by using normalized earnings or mid-cycle estimates instead of trailing actuals.

Why cyclical earnings distort the P/E

A steel mill, bank, or automaker’s earnings swing wildly with the business cycle. In an expansion, rising demand drives capacity utilization up, prices firm, and margins expand. Earnings soar 50–200% in 2–3 years. In a recession, demand collapses, utilization falls, prices crater, and many cyclical firms swing to losses.

The P/E ratio—price divided by earnings per share—mechanically responds to this earnings volatility. At peak earnings (the boom), earnings are at their highest, so the denominator is largest, and P/E looks lowest. At trough earnings (the recession), earnings are at their smallest, so the denominator is smallest, and P/E looks highest.

This inversion of intuition is the cardinal sin of cyclical valuation. A steel stock trading at $50 with trailing earnings of $10 per share (P/E of 5) looks like a bargain—cheaper than the market average of 15–18. But if that $10 is peak-cycle earnings and the stock’s true average earnings over a full cycle are $5, then the true P/E is really 10, not 5. The stock is not cheap; it is exactly fairly priced (or worse).

Conversely, the same stock in a trough—trading at $40 with earnings of $2 per share (P/E of 20)—looks absurdly expensive. But if mid-cycle earnings are $5, the true P/E is 8, a bargain. Investors who see a 20 P/E and assume the stock is overpriced miss the opportunity.

The peak trap and the trough trap

Peak-cycle trapping occurs at the height of a boom. Investors, dazzled by soaring earnings, see a low P/E and rush to buy. The stock price rises on momentum and FOMO (fear of missing out). But the boom is peaking; earnings growth is about to slow, then collapse. The stock, purchased at a seemingly cheap 6 P/E, is often purchased near its absolute top. When earnings fall 40% in the recession, the stock follows, and the buyer is left holding a 12–15 P/E on lower earnings—a loss.

Trough-cycle trapping is the mirror. Earnings are in free fall, the P/E looks sky-high (20+), and investors shun the stock thinking it is overvalued. But the trough is usually when the best opportunities hide. Early recoveries drive stock prices up long before earnings recover; buying at the trough can deliver 100%+ returns as earnings normalized and the P/E compresses back to historical mid-cycle levels.

A trader focused only on trailing P/E gets both bets backwards.

Normalized earnings: the fix

Normalized earnings are estimates of what earnings would be at a mid-cycle level of economic activity—neither boom nor bust. They smooth out cyclical spikes and troughs.

The most common approach is to average earnings over a full cycle, typically 7–10 years. A steel company might have earned $2, $5, $8, $10, $8, $4, $1, $3, $6, $7 per share over a 10-year cycle; the normalized earnings would be roughly $5.40 per share. Using this, a stock trading at $50 would have a normalized P/E of 9.3—a more faithful representation of value than a peak-cycle P/E of 5 or a trough-cycle P/E of 25.

Forward earnings estimates (next-year consensus from analysts) also help, but they lag normalized earnings because consensus often moves sluggishly. At a peak, forward estimates usually reflect a slowing of growth, not a collapse, so forward P/E is less misleading than trailing but still optimistic. At a trough, forward estimates are slow to turn bullish, so forward P/E is still inflated.

Management guidance on “normalized” or “run-rate” earnings (earnings excluding one-time items, adjusted for current activity levels) also informs analysts. But companies at peaks typically claim their peak earnings are repeatable, and at troughs, they claim the trough is transient. Management claims must be stress-tested against historical cycles.

A worked example: an airline

An airline stock, ABC Air:

• 2016 (trough): Earnings $1/share; stock $20; P/E = 20.
• 2017 (recovery): Earnings $3/share; stock $35; P/E = 11.7.
• 2018 (expansion): Earnings $6/share; stock $60; P/E = 10.
• 2019 (peak): Earnings $8/share; stock $75; P/E = 9.4.
• 2020 (collapse): Earnings −$3/share (loss); stock $25; P/E = undefined.

A naive investor in 2019 saw a 9.4 P/E and thought the airline was cheap relative to a 15 market average. They bought at $75, lured by the low multiple. But 2019 was peak earnings; the 10-year average earnings (2010–2019) for ABC Air were $4/share. A normalized P/E in 2019 would have been $75 ÷ $4 = 18.75—expensive, not cheap.

By 2020, the stock crashed. By 2021, as earnings recovered to $2, the stock bounced to $35. By 2022, earnings were $5, stock $55, normalized P/E = 11. Smart investors who knew the mid-cycle norm (around $4–5) and recognized the 2019 peak bought the 2020 trough and rode the 140% recovery.

Cyclical sectors and typical normalized P/E

Different cyclical sectors have different typical normalized P/E ranges:

These ranges are historical guides, not prescriptions. Structural changes (e.g., electric vehicles disrupting automakers, oil decline) can permanently lower normalized P/E. But they provide a check against peak-cycle euphoria and trough-cycle despair.

The limits of normalization

Normalized P/E is not foolproof. If a structural industry shift (e.g., digital disruption, regulatory change) is underway, historical cycle ranges are unreliable. Kodak’s normalized P/E looked reasonable in 2005–2008, but digital photography was permanently displacing film; historical norms were obsolete.

Also, mid-cycle earnings are themselves estimates. No two business cycles are identical. Analysts can differ sharply on what “normalized” means.

Still, for mature, non-disrupted cyclical businesses, normalized or forward earnings-based P/E ratios are vastly more informative than trailing P/E. They expose peak-cycle traps and reveal trough-cycle opportunities that a raw backward-looking multiple would hide.

See also