DCF: which cash flows you discount and why the terminal value decides everything

A DCF model values a company not by its market price but by the cash the business is able to hand over to its providers of capital across its entire working life: every future flow is cut by a factor that grows harsher the further away the year lies and the higher the required return. The sum of the discounted flows plus the terminal value gives the value of the enterprise, from which the value of a share is then derived. The output of such a model is not a price forecast but a set of assumptions fixed on paper: change the rate or the growth rate and the valuation will move further than it would after any revision to the revenue forecast.

What exactly is discounted: the flow and the rate must match

The main fork in the road comes at the very start. Free cash flow to the firm (FCFF) is the money left after taxes, capital expenditure and the change in working capital, but before payments to creditors; it belongs to all providers of capital and is discounted at the weighted average cost of capital (WACC). Free cash flow to equity (FCFE) takes interest and the movement of debt into account, belongs to shareholders alone and is discounted at the cost of equity.

A mismatch between the flow and the rate is the most common mistake in a beginner's model: a shareholders' flow discounted at WACC systematically overstates the valuation, because the effect of debt is counted again — in the flow and in the rate as well. The check is simple: if interest has been deducted from the flow, there should be no cost of debt in the denominator.

The forecast period: from the accounts, not from imagination

The forecast period is built along a chain: revenue — operating margin — income tax — capital expenditure — change in working capital. Every link must rest on the history of the specific company and on what management says, not on a desired trajectory. Historical values for margin, depreciation and capex are taken from the issuers' published financial statements, and the dates to which new inputs are tied come from the corporate events calendar.

The forecast horizon ends where the business reaches a steady state: a stable margin, capex in proportion to depreciation, growth that does not outpace the market. If the company is still growing in leaps by the end of the horizon, the horizon has been set too short — and everything that did not fit into the forecast will slide into the terminal value, where there is the least control.

The discount rate

WACC adds together the cost of equity and the cost of debt after the tax shield, weighting them by their market shares in the capital structure. The cost of equity is most often obtained through CAPM — the capital asset pricing model, in which beta multiplied by the equity market risk premium is added to the risk-free rate. The risk-free rate is taken from OFZ yields with a duration comparable to the valuation horizon; the cost of debt comes from the yield on the issuer's outstanding bonds on the debt market, not from the rate on an old loan.

Terminal value: the place where a valuation is made and broken

The lion's share of the final valuation falls not on the forecast years but on the terminal part. It is usually calculated through the Gordon model — the constant growth model: the flow of the first post-forecast year is divided by the difference between the discount rate and the sustainable growth rate. A disciplining constraint follows from this: the growth rate cannot exceed the long-term growth of the economy, otherwise the company ultimately becomes bigger than the market, and as the rate and the growth rate converge the denominator blows the valuation up.

The alternative is an exit multiple: the terminal year is valued at a comparable EV/EBITDA and discounted back to the present day. This method shows more honestly that you are leaning on a market norm, but it inherits that norm's errors.

From the value of the business to the value of a share

The discounted sum of FCFF flows gives the enterprise value. Net debt and the non-controlling interest are subtracted from it, non-operating assets — excess cash, non-core real estate, financial investments — are added, and the result is divided by the number of shares outstanding. The figure obtained is compared with current share prices. For dividend stories a cross-check is useful: the capacity to pay that is built into the model should not contradict the actual dividend practice.

Where DCF does not work

For banks and insurers free cash flow is poorly defined: for them debt is the raw material of operating activity, not a source of financing. There a dividend-based valuation — the dividend discount model — or a valuation based on excess return on capital is more appropriate. Cyclical companies require the flow to be normalised over a full cycle rather than extrapolated from a good year. For a business without predictable flows, DCF gives an illusion of precision: a long formula adds no knowledge about the future.

This is exactly the case that model risk deals with — the risk in the construction of the model itself. A useful antidote is the reverse DCF: take the market price as a given and work out what growth and margin are built into it. That question can be tested, unlike the question of "what the company is really worth".

A separate point should be stated honestly: this article does not give specific values for rates, betas and multiples — they change every trading day, and they should be plugged in from up-to-date data, not from the text of an article.