Using the Dividend Discount Model for Cyclical Companies

The dividend discount model for cyclical companies requires special care because dividends swing wildly with the business cycle, making the traditional assumption of stable perpetual growth unrealistic and often misleading.

Why Standard DDM Breaks Down for Cyclicals

A traditional dividend discount model assumes dividends grow at a constant rate into perpetuity. That assumption works well for utilities, consumer staples, and other non-cyclical firms. But for companies tied to commodity prices, industrial capex cycles, or credit availability—banks, mining, chemicals, automakers, shipping—dividends are profoundly pro-cyclical.

A bank might pay $2 per share in dividends during a boom, then cut to $0.50 during a crisis as loan losses mount. A mining company’s dividend swells when copper prices spike, then vanishes when they crash. If you mechanically fit a 4% growth rate to last year’s $2 dividend, you’re implicitly assuming the cycle has ended and boom-time dividends are the new normal. That’s the opposite of cautious analysis.

Normalizing to Mid-Cycle Earnings

The standard correction is to estimate “normalized” or “through-cycle” earnings—what the company would earn at average cyclical conditions, not peak or trough. This discipline forces you to make an explicit forecast about where in the cycle the company sits.

Example: A bank with volatile earnings

Suppose a regional bank shows:

• Peak cycle earnings: $4.00 per share (historically, during low-loss years)
• Trough cycle earnings: $0.50 per share (during credit downturns)
• Current earnings: $3.50 per share (near peak, as credit quality is strong)

A naive analyst might assume 4% growth on $3.50 dividends, yielding a value of $3.50 × 1.04 / (0.10 − 0.04) = $60.67 per share. But this assumes peak earnings persist indefinitely. A more defensible approach:

1. Estimate normalized earnings at mid-cycle: $2.25 per share (the geometric mean, roughly).
2. Estimate a normalized payout ratio (say, 50% during a normal year): $1.125 per share in normalized dividend.
3. Assume modest 2% long-term growth (tied to GDP): $1.125 × 1.02 / 0.06 = $19.12 per share.

The gap is enormous. If the bank currently pays $1.75 (near peak payout), the normalized-earnings approach suggests it’s overvalued unless you believe the peak is permanent.

Two-Stage Cyclical Models

Another approach uses explicit forecast periods for the near-term cycle, then a stable terminal value. This is useful when you have a formed view on where the cycle is headed.

Example: An automaker in early cycle recovery

• Years 1–3: Near-term forecast (industry gradually recovering)
  • Year 1 dividend: $1.50 (trough-ish)
  • Year 2 dividend: $2.50 (mid-cycle)
  • Year 3 dividend: $3.00 (approaching peak)
• Years 4+: Stable state
  • Terminal dividend: $2.25 (normalized, mid-cycle level)
  • Growth rate: 2%
  • Discount rate: 10%

You discount each of years 1–3 explicitly, then calculate a terminal value using the perpetuity formula on year 4 onward. This avoids the trap of either extrapolating trough or peak indefinitely.

The benefit: you’re transparent about your cyclical view. If the cycle evolves differently than expected, you can re-run the model with new explicit-period forecasts. If the industry enters recession instead of recovery, you adjust the Year 2–3 dividends downward.

Adjusting Payout Ratios Over the Cycle

Cyclical companies often maintain more conservative payout ratios than their current earnings would suggest. A mining company earning $5 per share during a commodity boom might pay only $1.50, retaining cash to preserve financial flexibility for the downturn. Alternatively, during a trough, it might cut dividends despite guidance that recovery is near.

When building a normalized DDM:

1. Estimate normalized earnings (mid-cycle).
2. Research management’s stated target payout ratio or infer it from historical peaks and troughs.
3. Use the normalized payout ratio, not the current one. If current earnings are $4 and the company pays $1.50, that’s a 37.5% payout. If normalized earnings are $2.25, a comparable payout would be $2.25 × 0.375 = $0.84, not $1.50.

This step is critical because it decouples dividend from the cyclical earnings level.

The Role of Dividends Versus Buybacks

Some cyclical firms use buybacks to return capital in boom years, reserving dividends for years when earnings decline. This makes dividend-based valuation incomplete. A full excess return dividend model or free-cash-flow framework might be more appropriate. But if a cyclical firm has a longstanding commitment to maintaining or growing its stated dividend rate even through cycles, that promise is often more stable than reported earnings and can anchor the valuation.

Scrutinize the dividend policy language in 10-Ks and guidance. If management commits to “dividends may be cut in a recession” or “we retain flexibility,” treat the current dividend as cyclical. If they say “we have not cut our dividend in 30 years,” that’s signal of a managed, normalized payout.

Stress-Testing Cyclical Assumptions

Always run sensitivity analysis on cyclical valuations with wider ranges than you’d use for utilities or consumer staples. Test:

• What if normalized earnings are 10% lower than your estimate? (Cyclical margins compress permanently.)
• What if the payout ratio falls to 40% instead of 50%? (Financial stress forces conservatism.)
• What if growth is 0% instead of 2%? (The cycle flattens; no secular growth.)

If reasonable changes in these assumptions produce value swings of ±30% or more, the model is telling you that cyclical visibility is low and your valuation contains significant judgment risk.

Red Flags in Cyclical Valuation

Extrapolating peak earnings to perpetuity: The most common mistake. If the last five years have been boom, analysts often model 5% growth from an elevated earnings base. That’s rarely sustainable when a trough inevitably arrives.

Ignoring balance sheet stress during downturns: Some cyclicals maintain dividends even as profitability craters, drawing on cash reserves or increasing debt. This is unsustainable. Normalized dividend should reflect what’s payable from normalized free cash flow, not just accounting earnings.

Misaligning the cycle timeline: If you forecast a two-year recovery but the cycle actually takes five years, your intermediate-period dividend forecasts are wrong. Be skeptical of precise timing claims; instead, bracket with pessimistic and optimistic scenario valuations.

When to Use DDM for Cyclicals (And When Not To)

DDM works best for cyclical companies with a long dividend history and a clearly stated commitment to maintaining or growing it through cycles. Diversified industrials, integrated energy majors, and large banks often fit this profile. It works poorly for small-cap cyclicals with erratic payout histories, or for firms that eliminate dividends during recessions.

For true commodities cycles (oil, metals) where dividends depend directly on commodity prices, consider using scenario analysis: “if oil averages $70/bbl over a cycle, normalized dividend is $X; if $50/bbl, it’s $Y.” This is more honest than fitting a growth rate.

See also