How to Build a DCF Model: Build Order, Terminal Value and the Sanity Checks

Unlevered free cash flow: the build

Start from EBIT, subtract taxes calculated on EBIT rather than on earnings before tax, add back depreciation and amortisation, subtract capital expenditure, and subtract the increase in net working capital. The result is the cash the business generates before any financing decision.

A worked case: EBIT of $400m, a 25% tax rate, D&A of $120m, capex of $150m and a $30m increase in working capital. Unlevered free cash flow is $400m less $100m of tax, plus $120m, less $150m, less $30m, which comes to $240m.

The reason tax is calculated on EBIT rather than on EBT is that unlevered cash flow must be independent of the capital structure. Taxing EBT would import the interest tax shield into the cash flows, and that shield is already accounted for in the after-tax cost of debt inside WACC. Doing both double-counts it.

WACC: build it component by component

The formula is the equity weight times the cost of equity plus the debt weight times the after-tax cost of debt, using market values throughout. What matters in a review is not the formula but whether each component was derived or assumed.

Cost of equity comes from CAPM: the risk-free rate plus levered beta times the equity risk premium. Beta should generally be taken from comparable companies — unlevered to strip out each peer’s capital structure, averaged, then relevered at your target’s own structure. A thinly traded company’s own historical beta is not a defensible input.

A worked case: a 4.2% risk-free rate, a relevered beta of 1.2 and a 5.5% equity risk premium give a cost of equity of 10.8%. At 70% equity and 30% debt by market value, with a 6.0% pre-tax cost of debt and a 25% tax rate, WACC is 0.7 times 10.8% plus 0.3 times 6.0% times 0.75, or 8.9%.

• Market values of debt and equity, never book values.
• Peer-derived beta, unlevered and relevered — show the working in the model.
• Cost of debt after tax. Forgetting the (1 − t) factor overstates WACC and understates the valuation.
• One WACC per risk profile. A conglomerate with genuinely different divisional risk needs more than one.

Terminal value: build both methods, then argue

Terminal value is typically 60% to 75% of total enterprise value in a ten-year model, which means the DCF is mostly a statement about the terminal assumption. Treating it as a footnote is the most common way a DCF loses credibility.

The perpetuity growth method: final-year free cash flow times one plus g, divided by WACC less g. With $500m of final-year cash flow, 2.5% growth and a 9% WACC, that is $512.5m divided by 0.065, or $7.88bn.

The exit multiple method: final-year EBITDA times a terminal EV/EBITDA multiple anchored on where comparables trade today.

Build both, then cross-check. Back out the growth rate implied by your exit multiple; if it comes to 5%, the multiple is too high and you should either defend that explicitly or lower it. Back out the exit multiple implied by your perpetuity growth rate; if it implies a multiple far above where the sector trades, the growth rate is doing work it cannot support. This reconciliation is what a good reviewer looks for and most models omit.

Discounting, and the mid-year convention

Discount each forecast year’s free cash flow at WACC, and discount the terminal value back from the end of the forecast period. Forgetting to discount terminal value is a mechanical error that inflates the valuation by a large margin and is more common than it should be.

The mid-year convention treats cash flows as arriving evenly through the year rather than in a lump on the final day, using discount periods of 0.5, 1.5, 2.5 and so on. It raises the valuation modestly and is more realistic for most operating businesses.

If you apply it, apply it consistently. The frequent error is mid-year discounting the forecast cash flows while discounting terminal value on a full-year basis, which quietly mixes two conventions.

The equity bridge

Summing the discounted cash flows and discounted terminal value gives enterprise value. To get to a share price: subtract total debt, subtract preferred stock, subtract minority interest, add cash and equivalents, then divide by the fully diluted share count.

Use diluted shares, calculated with the treasury stock method for in-the-money options. Using basic shares overstates the per-share value, and it is the kind of detail a reviewer checks precisely because it is easy to skip.

The output is a range, not a number

A DCF that produces a single share price is not finished. The output should be a sensitivity table across the two assumptions that drive the answer — usually WACC on one axis and terminal growth rate (or exit multiple) on the other — showing the implied value across a plausible range of each.

This is not hedging. It is the honest presentation of a model whose output moves materially on inputs that cannot be observed. A reviewer who sees a point estimate immediately asks what happens if the discount rate is 50 basis points higher; the sensitivity table answers that before the question is asked, and it demonstrates that you know where the model is fragile.

Present the DCF range alongside comparable companies and precedent transactions on a football field. A DCF sitting entirely outside the market-based ranges is not necessarily wrong, but it needs an explicit explanation, and it is better to give that explanation yourself than to be asked for it.

When a DCF is the wrong tool

A DCF requires forecastable cash flows. Where they are not forecastable, the model produces precision without accuracy, and using one signals a lack of judgement rather than rigour.

Banks and insurers are the clearest case: their capital structure is the business, so an unlevered DCF is conceptually meaningless. They are valued on a dividend discount model or price-to-book instead. Early-stage companies with no earnings and deeply cyclical businesses near a turning point are the other two common cases where a DCF is the wrong instrument.