Discounted Cash Flow (DCF)

The core formula

A DCF computes enterprise value (or equity value) as:

Value = Σₜ FCFₜ / (1 + r)ᵗ  +  TV / (1 + r)ᴺ

where TV = FCFₙ × (1 + g) / (r − g)    [Gordon growth]
or    TV = FCFₙ × multiple             [exit multiple]

The first term sums the present value of explicit forecast cash flows over N years (typically 5-10). The second term — the terminal value — captures everything from year N+1 to infinity. Both pieces are discounted to today at rate r.

The three inputs you must defend

1. Free cash flow. The cash a business generates after reinvesting enough to maintain operations.
   FCFF = EBIT × (1 − tax) + Depreciation − CapEx − ΔWorking Capital.
   FCFE = FCFF − Interest × (1 − tax) + Net Borrowing.
2. Discount rate. WACC for FCFF, cost of equity for FCFE.
   WACC = (E/V) × r_E + (D/V) × r_D × (1 − tax).
3. Terminal value. Either Gordon growth (perpetual constant growth) or exit multiple (apply a peer-group multiple to year-N EBITDA or earnings).

Worked example: 5-year FAANG-style DCF

Imagine a hypothetical mature tech company with these starting assumptions:

Year 0 revenue:   $100B
Year 0 FCF:        $25B (25% FCF margin)
Revenue growth:    10% years 1–3, 7% years 4–5
FCF margin:        25% steady
WACC:              9%
Terminal growth:   3% (≈ long-run nominal GDP)

Project the explicit cash flows:

Year	Revenue ($B)	FCF ($B)	Discount factor (1.09⁻ᵗ)	PV of FCF ($B)
1	110.0	27.50	0.917	25.23
2	121.0	30.25	0.842	25.47
3	133.1	33.28	0.772	25.69
4	142.4	35.61	0.708	25.21
5	152.4	38.10	0.650	24.77

Sum of PV(FCF) years 1-5 = $126.4B.

Terminal value at end of year 5:

TV₅ = FCF₅ × (1 + g) / (r − g)
    = 38.10 × 1.03 / (0.09 − 0.03)
    = 39.24 / 0.06
    = $654.0B

PV of TV = TV₅ × 1.09⁻⁵ = 654.0 × 0.650 = $425.1B

Enterprise value = 126.4 + 425.1 = $551.5B.

Notice: terminal value contributes 425.1 / 551.5 = 77% of total EV, even though it represents cash flows that haven't happened yet and rest on assumptions you'll never get to verify within your career. The 5-year explicit forecast — the part you actually built spreadsheets for — is responsible for less than a quarter of the answer.

Sensitivity to terminal value assumptions

	g = 2%	g = 3%	g = 4%
WACC = 8%	$595B	$671B	$789B
WACC = 9%	$498B	$552B	$629B
WACC = 10%	$427B	$466B	$521B

A 1 percentage point change in either WACC or terminal growth moves the valuation by 12-25%. This is why bankers run sensitivity tables, not point estimates — and why you should treat any single-number DCF result with deep skepticism.

DCF vs other valuation methods

Method	What it computes	Strength	Weakness	Best used for
DCF	Intrinsic value from projected cash flows	Theoretically pure; works for unique businesses	Garbage in, garbage out; terminal value dominates	Stable mature companies with predictable cash flows
Comparable Companies (trading multiples)	Apply peer-group EV/EBITDA, P/E, EV/Sales	Reflects current market sentiment; quick	Inherits market mispricing; "comparable" is subjective	Quick triangulation; sectors with many similar listed peers
Precedent Transactions	Apply multiples paid in past M&A deals	Captures control premium; reflects strategic value	Old deals reflect old market conditions; small samples	M&A advisory; pricing acquisitions
Sum-of-the-parts (SOTP)	Value each business segment separately, sum	Captures conglomerate discount/premium	Requires segment financials; subjective allocation	Multi-business holding companies
Liquidation value	Net realizable value of assets minus liabilities	Hard floor on equity value	Ignores going-concern value	Distressed situations; deep value bottom-fishing
Real options (Black-Scholes valuation)	Value of optionality embedded in projects	Captures upside in uncertain investments	Mathematically tractable but assumption-heavy	Mining, R&D, expansion options
Replacement cost	What it would cost to rebuild the business from scratch	Cross-checks DCF on asset-heavy businesses	Doesn't capture intangible value (brand, network effects)	Industrial assets, real estate, infrastructure

Damodaran's data on professional valuation work: most pitch decks include all three of DCF, comps, and precedents. A "football field" chart shows the valuation range from each method, with the recommendation typically falling in the overlap zone.

Exit multiple terminal value: an alternative

Instead of Gordon growth, some practitioners apply an exit multiple to year-N EBITDA:

TV = EBITDA_N × (peer-group EV/EBITDA multiple)

This anchors terminal value to current market multiples instead of perpetuity assumptions. The implicit growth rate that justifies the multiple should be cross-checked against Gordon growth — they should converge. If your exit multiple implies 8% perpetual growth, you've either picked a bubble multiple or extended the high-growth phase past where it's defensible.

Where DCFs go wrong

• Hockey-stick forecasts. Analysts pencil in flat margins for 4 years, then expand them in year 5 to make the model work. Smell test: if year-5 margins exceed any peer's, you're forecasting industry-leading economics that haven't been earned.
• Terminal value engineered to a target. Pick a price first, then back into the terminal growth that produces it. This is rampant in sell-side research.
• Inconsistent inflation assumptions. Nominal cash flows discounted at real rate, or vice versa. The 200-300 bp discrepancy can swing valuation by 30%.
• Wrong free cash flow definition. Forgetting to add back stock-based compensation in tech, or subtracting it as cash expense. Modern Damodaran practice: treat SBC as cash and add restricted shares to share count.
• Ignoring working capital growth. Fast-growing companies fund inventory and receivables with cash. Forgetting to subtract ΔWC overstates FCF in growth phases.
• Using levered β unlevered. Need β_unlevered when WACC is computed; need β_levered for cost of equity. Mixing them produces an artificially low or high WACC.
• Discount rate that doesn't match cash flow currency. USD cash flows need a USD discount rate; EUR cash flows need EUR. Cross-currency forecasts require translation at forward exchange rates.

When DCF is the wrong tool

• Pre-revenue or early-stage companies. Cash flows are 7+ years out; terminal value is essentially 100% of value; assumptions are sci-fi. Use venture-style scenario analysis or comparable startups instead.
• Cyclical businesses at peak/trough. Project off-cycle averages, not current numbers. A steel company at peak earnings, valued via DCF using current FCF, looks artificially cheap.
• Companies with optionality. Mining concessions, biotech pipelines, oil reserves. Real options valuation captures upside DCF flattens.
• Distressed companies. Going-concern value is uncertain; liquidation value or sum-of-parts is more relevant.
• Banks and insurers. Free cash flow is hard to define when the business itself is intermediating cash. Use dividend discount or residual income instead.

Sanity checks every DCF should pass

• Implied EV/EBITDA at end of forecast. Compute terminal EV ÷ year-N EBITDA. Should land near peer multiples — far above and you've engineered a mispricing.
• Terminal value as % of total EV. Above 90%? Forecast horizon is too short. Below 30%? Probably forecasting too aggressively in years 1-N.
• Implied perpetual revenue growth vs GDP. Nominal terminal growth must be ≤ long-run nominal GDP for the firm not to swallow the economy.
• Implied ROIC vs WACC in steady state. If terminal ROIC >> WACC and the firm has no moat justification, value-creation can't persist forever.
• Cross-check vs trading multiples. If your DCF EV is 3× the comp-implied EV, you're either right about something the market is missing or wrong about something the market knows.