Valuing Cyclical Stocks with Multiples
Cyclical stocks—companies whose earnings swing wildly with economic booms and busts—demand a fundamentally different approach to valuation than stable, predictable firms. A standard P/E ratio snapshot tells you almost nothing when earnings are at a trough or a peak. This chapter walks you through the discipline of normalizing earnings, choosing the right valuation periods, and recognizing when multiples hide dangerous assumptions about mean reversion.
Quick definition: Normalized earnings are a company's estimated average earnings power across a full business cycle, adjusted for cyclical peaks and troughs—the basis on which cyclical multiples should be computed.
Key Takeaways
- Never use trailing earnings alone for cyclical companies; instead, normalize earnings across a full cycle to arrive at a fair valuation metric.
- Peak and trough earnings require separate analysis—understand whether the stock is being valued on worst-case or best-case assumptions.
- EV/EBITDA and EV/Sales often more reliable than P/E for cyclical names, since they ignore financing and tax distortions that swing with the cycle.
- Peer comparisons work only within industry segments—an auto manufacturer and a chemical company cycle differently; compare within your lane.
- Multiple compression is the primary cyclical risk—even if earnings recover, the multiple applied to those earnings may decline as the cycle matures.
- Mean reversion is not guaranteed—macroeconomic shocks, structural disruption, and supply-side changes can permanently alter a cyclical's normalized earnings power.
What Makes Cyclical Valuation Different
Imagine a steel manufacturer reporting $5 billion in EBITDA during a commodity boom year, then $1 billion three years later when demand collapses. A P/E of 8x on peak earnings looks cheap; a P/E of 25x on trough earnings looks expensive—yet they describe the same business. The solution is to estimate the company's normalized earnings power: the annual EBITDA or free cash flow it would earn in a mid-cycle environment, neither a boom nor a recession.
Cyclical businesses span auto, chemicals, steel, semiconductors, mining, construction, real estate, and energy. Each has its own cycle length and amplitude. Semiconductors swing every three to five years; real estate cycles last a decade; precious metals can gyrate on currency and central bank policy shifts. The key is understanding where in the cycle your valuation snapshot falls.
Normalized Earnings: The Starting Point
Normalized earnings are not an official accounting figure—they're an estimate you must make. Here's the standard approach:
Step 1: Identify the cycle length. How many years has it taken historically for earnings to move from trough to peak and back? If you can't find historical data, proxy using industry reports or analyst consensus on cycle timing.
Step 2: Collect multi-year earnings. Gather trailing EBITDA (or free cash flow) for the past 5–10 years, covering at least one full cycle.
Step 3: Estimate the normalized midpoint. Take a weighted average of mid-cycle earnings—years when the company was neither booming nor in deep recession. If that's subjective, use the median earnings across the full period, or a three-year trailing average centered on a neutral year.
Step 4: Apply the normalized earnings to current multiples. Once you have a normalized annual EBITDA figure (say, $3 billion), apply an appropriate peer multiple (say, 8x EV/EBITDA for your steel cohort) to get an enterprise value, then subtract net debt to find equity value.
For example, if US Steel has seen EBITDA range from $500M (recession) to $4B (boom), a reasonable normalized estimate might be $1.8B. At an 8x EV/EBITDA multiple (median for the steel group), the enterprise value is $14.4B. Subtract net debt and you have an equity value anchor.
Why Standard P/E Fails for Cyclicals
The P/E ratio amplifies the cyclical problem because net income is net—it's after interest, taxes, and all the distortions that swing with the cycle. When a cyclical business hits a downturn:
- Interest coverage tightens, sometimes triggering higher debt costs or covenant violations.
- Tax rates swing (loss carryforwards lower the tax rate in down years; normalcy raises it).
- Reported earnings can swing into loss territory, making P/E undefined or wildly negative.
EV/EBITDA and EV/Sales strip out these financing and tax effects, making them more comparable across cyclical peaks and troughs. They also let you see the true operating economics:
- EV/EBITDA = Enterprise Value ÷ Operating Earnings, before interest, taxes, depreciation. Stable across the cycle.
- EV/Sales = Enterprise Value ÷ Total Revenue. Immune to margin compression, but ignores profitability entirely.
For a steel company in distress, EV/Sales might be 0.4x; at peak, 0.8x. EV/EBITDA might be 6x in trough (when EBITDA is low) and 5x at peak (when EBITDA is high). These moves are more modest than what you'd see in P/E, which might be undefined or negative in a trough and 20x+ at peak.
The Normalized EBITDA Multiple
Once you settle on a normalized EBITDA figure, the next step is choosing the right multiple. Here's where peer analysis becomes critical.
Within-industry peers are essential. An automotive supplier that makes parts for Tesla has a very different cycle and margin profile than a diversified machinery manufacturer. Know which peer subset is truly comparable.
Adjust for leverage. In a normalcy scenario, different cyclicals carry different debt levels. A utility with stable cash flow might be financed at 60% debt-to-capital; a cyclical mining company, 30%. The EV/EBITDA multiple reflects both business risk and financial risk. If your target company is underleveraged relative to peers, it may deserve a higher multiple (less financial risk); if it's overleveraged, a lower multiple (more risk of distress).
Back-test the normalized multiple. Look at 5–10 years of history. In the last three mid-cycle years, what was the actual EV/EBITDA? Use that as a guide, not a guarantee. If the multiple has compressed structurally (e.g., due to excess industry capacity), that's a red flag for structural, not cyclical, decline.
Peak and Trough Valuation Scenarios
Professional analysts often build three scenarios: trough, normalized, and peak. This forces clarity on downside and upside.
- Trough scenario: Assume the company is at the bottom of the cycle. Estimate trough-year EBITDA (often 30–50% of peak), apply a higher multiple (because risk is high and the multiple is depressed), and calculate an enterprise value that represents maximum distress.
- Normalized scenario: Use the normalized EBITDA and a median peer multiple. This is your fair-value estimate.
- Peak scenario: Assume peak-cycle EBITDA, apply a peer multiple (often lower at peak due to cyclical euphoria leading to overleverage), and calculate upside.
For a $20B industrial company:
- Trough: $800M EBITDA × 7x = $5.6B EV = $2B equity (if net debt is $3.6B).
- Normalized: $1,500M EBITDA × 8x = $12B EV = $8B equity.
- Peak: $2,200M EBITDA × 6.5x = $14.3B EV = $10.3B equity.
The normalized case should be your base case. The trough and peak cases tell you the range of outcomes if your cycle-timing assumption is wrong.
A Mermaid View of Cyclical Valuation Logic
Practical Adjustments for Leverage
A cyclical company's debt-to-EBITDA ratio swings. At peak, debt might be just 1.5x EBITDA (because EBITDA is soaring); at trough, 4x or 5x. When valuing, anchor on what debt-to-EBITDA would be in a normalized environment, not where it sits today.
If the normalized debt-to-EBITDA is 2.5x and current debt is $4B, then normalized EBITDA is $1.6B, not the trailing $2.4B (if the company is in a peak). This adjustment can dramatically change your valuation.
Example: A mining company has $6B net debt and trailing EBITDA of $2B (at peak commodity prices). Peers trade at 5x normalized EV/EBITDA. But historical data suggests normalized EBITDA is closer to $1.2B. At 5x, the EV is $6B; subtract net debt of $6B and you have $0 in equity value—a red flag. The market price reflects an assumption that the peak holds; fundamental analysis warns you it won't.
When Multiples Compress Despite Earnings Recovery
One of the cruelest traps in cyclical investing is multiple compression: earnings recover, but investors won't pay the same multiple they once did.
This happens when:
- Structural oversupply persists. The industry added too much capacity in the boom; even as the cycle recovers, supply-demand balance takes years to normalize.
- Return on capital collapsed. Once-profitable cyclicals may recover volume but at lower prices or margins, yielding ROIC below the cost of capital—the market reprices them lower.
- Competitive intensity rose. A regional oligopoly fragmented into a hypercompetitive market; the cycle can't reverse that.
- Technological disruption eroded margins. Falling costs for a substitute product mean the cyclical can never return to its old profitability.
The antidote is to look beyond the current cycle. Ask: will the normalized EBITDA multiple—not today's depressed multiple—re-rate back to historical levels, or has it structurally declined? If the latter, the upside from cycle recovery is limited, even if earnings do bounce.
Real-World Examples
US Steel in 2016–2020: US Steel reported peak EBITDA of $3.5B (2018), then trough EBITDA of negative $200M (2020, covid). Historical normalized EBITDA was closer to $1.2B. By 2022, earnings had recovered to ~$2B, but the EV/EBITDA multiple was stuck at 4–5x, versus 7–8x ten years earlier. Why? Industry overcapacity, slowing US auto production, and cheaper imports permanently eroded margins. Investors repriced the multiple lower. Even with earnings recovery, equity returns lagged.
Weyerhaeuser (timber/REIT) 2020–2023: Logged distressed lows in March 2020 at a 3.5x EV/EBITDA, with normalized EBITDA estimated at ~$1.3B. Lumber prices soared in 2021–2023, pushing EBITDA toward $2.5B. The multiple re-rated to 7x. The combination of earnings recovery + multiple expansion delivered a 3x+ equity return—a textbook cyclical rebound.
Samsung Electronics 2022–2024: The chip cycle peaked in 2021, then trough in 2023 (EBITDA fell 60%). Normalized EBITDA across a full cycle (memory + logic) was estimated at $20B. By early 2024, EBITDA had begun recovering from $8B trough toward $16–17B. The multiple was still depressed at 4.5x EV/EBITDA. As the cycle normalized, multiples typically re-rate to 6–7x; the combination offered upside, but it depended on mean reversion.
Common Mistakes
1. Using peak earnings as normalized earnings. Peak years feel like "normal" to recent investors. They're not. A mining company that reports $2B EBITDA in a commodity supercycle should not be valued as if it will earn $2B forever. Normalize downward.
2. Ignoring leverage through the cycle. A company that finances peak earnings at 60% debt will struggle at trough. When it normalizes, the interest burden may be too high, forcing asset sales or equity dilution. Stress-test debt through a trough scenario.
3. Assuming the cycle length has changed. Every few years, someone claims "this time is different" because the cycle is longer. Often it's not. Refer to the full 10–20 year history, not just the last boom.
4. Mixing in structural decline with cyclical recovery. Some industries are cycling and dying at the same time (e.g., coal power plants during an energy transition). Separate the cyclical swing from the secular trend. A recovering coal company is not a buy if coal demand is structurally declining.
5. Over-weighting the normalized multiple. Even within a peer group, the multiple can shift if leverage, ROIC, or growth rates diverge. A peer might justify a 7x multiple; your company might deserve 6x if it has worse capital allocation or weaker pricing power.
FAQ
Q: Is there a rule of thumb for normalized earnings periods?
A: Most practitioners use 5–10 years for a full cycle, with the median or a weighted mid-cycle average as the normalized figure. For very long cycles (real estate, utilities), 10–15 years may be required. If the company is less than 5 years old or the industry is in transition, normalized earnings are harder to estimate and carry higher uncertainty.
Q: Should I ever use a cyclical's trough multiples as fair value?
A: No, not as your primary anchor. Trough multiples (like 10x+ EV/EBITDA for distressed cyclicals) reflect maximum fear and financial risk, not fair value. Use them to understand downside. Fair value is the normalized case.
Q: How do I know if a cyclical is at a trough or peak right now?
A: Compare trailing EBITDA to the 5–10 year average. If trailing is 40% below average and falling, you're likely approaching trough. If it's 50% above average and growth is moderating, you're likely in late peak. Cross-check with industry-published capacity utilization, inventory levels, and order flows.
Q: Can I use dividend yield to value a cyclical?
A: With caution. Many cyclicals cut dividends in troughs (or go to zero), so trailing yield is often deceptive. Forward yield—based on a normalized dividend—can work, but only if the company has a clear policy of maintaining or growing the payout across cycles. Most cyclicals don't; they return excess cash opportunistically at peaks.
Q: How do I adjust for FX exposure in a cyclical with international operations?
A: For a mining company that exports, include currency exposure in your trough-case modeling. A weaker US dollar helped exporters in 2023; a stronger dollar in 2024 hurt them. Normalize currency at a 3–5 year average rate, or model upside/downside currency scenarios separately.
Q: Should I weight the normalized multiple toward historical highs or lows?
A: Weight toward the mid-cycle or normalized environment. If peers have traded 6–8x EV/EBITDA over the full cycle, 7x is your anchor, not 8x (which implies the cycle is permanently shifted higher). Use 6x if you believe structural headwinds will keep multiples depressed.
Related Concepts
- Industry cycles and business cycles – Understanding where an industry sits in its boom-bust rhythm and how macroeconomic shocks affect cyclicals differently.
- Debt and leverage through the cycle – How to stress-test a balance sheet across trough scenarios and estimate distress probability.
- ROIC and mean reversion – Whether a cyclical's high returns in peak years will persist or normalize downward.
- Multiple expansion and contraction – The risk that even if earnings recover, the valuation multiple compresses, capping upside.
- Peer-set construction for cyclicals – Why comparing a regional steel mill to a global integrated producer distorts valuation.
- Scenario analysis and sensitivity – Building trough, normalized, and peak cases to quantify range of outcomes.
Summary
Cyclical stocks demand normalized earnings multiples, not snapshots of current earnings. Identify the cycle length, estimate normalized EBITDA or free cash flow, apply peer multiples, and build trough-normalized-peak scenarios to quantify risk and upside. Watch for multiple compression—earnings recovery does not guarantee equity returns if the market reprices the business lower. Adjust for leverage, separate cyclical recovery from structural decline, and remember that mean reversion is a tendency, not a law. The disciplined cyclical investor asks: is this company cheap relative to normalized earnings power, or cheap because the cycle will never recover?