Fade Periods and Convergence to Competitive Equilibrium

A pharmaceutical company earns a 40 percent return on invested capital (ROIC) while its blockbuster drug is under patent. The day the patent expires, generic competitors flood in and that 40 percent starts falling toward whatever the industry's normal, competitive return looks like — maybe 10 percent. The question a valuation model has to answer isn't just "how high is the return today," it's "how many years does it take to fall, and along what path." That transition is the fade period.

Think of a hot new restaurant that opens to two-hour waits and premium prices. Word spreads, imitators open across the street, the novelty wears off, and within a few years the wait time and pricing converge to whatever's normal for a good restaurant in that neighborhood. Nobody expects the two-hour wait to last forever, and nobody should model it that way — the only real questions are how long the fade takes and how far it ultimately reverts.

Modeling the fade

A simple linear fade takes a firm's excess return over its cost of capital and shrinks it toward zero over $N$ years:

$$\text{ROIC}_t = r + \left(\text{ROIC}_0 - r\right)\times\left(1 - \frac{t}{N}\right), \quad 0 \le t \le N$$

Here $\text{ROIC}_t$ is the return on invested capital in year $t$ of the fade, $\text{ROIC}_0$ is today's abnormal return, $r$ is the firm's cost of capital (the competitive, no-abnormal-profit benchmark), and $N$ is the number of years the fade is assumed to take. In words: the gap between today's return and the "normal" return shrinks in a straight line, hitting exactly the cost of capital at year $N$ and staying there in perpetuity — because once returns equal the cost of capital, growth itself creates no additional value, which is what makes it a stable terminal assumption.

Worked example: a 10-year fade

A software company earns ROIC of 25 percent today against a cost of capital of 9 percent, and the analyst assumes an 8-year fade to competitive equilibrium as enterprise customers gain bargaining power and switching costs erode.

At year 4: $\text{ROIC}_4 = 0.09 + (0.25 - 0.09)\times(1 - 4/8) = 0.09 + 0.16\times0.5 = 0.17$, or 17 percent.

At year 8: $\text{ROIC}_8 = 0.09 + 0.16\times(1 - 1) = 0.09$, exactly the cost of capital — abnormal profitability has fully faded, right on schedule.

Worked example: fade period length swings terminal value

Two analysts value the same firm, agreeing on every assumption except fade length: one uses a 5-year fade, the other a 15-year fade, from the same 25 percent ROIC down to the same 9 percent cost of capital. The 15-year analyst has the firm earning abnormally high returns on a much larger base of reinvested capital for three times as long. Running representative cash flow assumptions through both, the 15-year fade typically produces a terminal value 30 to 50 percent higher than the 5-year fade — a single, largely qualitative judgment about "how durable is this moat" can move the valuation by nearly half, more than most of the explicit forecast years combined.

What this means in practice

Because terminal value routinely accounts for 60 to 80 percent of a DCF's total value, the fade-period assumption is one of the highest-leverage judgment calls in the entire model — far more consequential than fine-tuning next year's revenue growth. Analysts anchor fade length to observable evidence: patent cliffs, historical margin-reversion speed in the industry, and the strength of the company's actual moat (network effects, switching costs, regulatory protection) rather than picking a round number like 10 years by habit.

Key terms

• ROIC — return on invested capital; profit generated per dollar of capital employed in the business.
• Fade period — the number of years assumed for abnormal returns to converge to the cost of capital.
• Competitive equilibrium — the state where ROIC equals the cost of capital and growth creates no additional value.
• Terminal value — the value of all cash flows beyond the explicit forecast horizon, highly sensitive to fade assumptions.