Walk me through a DCF: the valuation method, explained with a worked example

The idea in one paragraph

A business is worth the cash it will give its owners, adjusted for the fact that cash in the future is worth less than cash today and that uncertain cash is worth less than certain cash. A discounted cash flow valuation makes that sentence into arithmetic. You forecast the cash the operations will generate for a few years, estimate what the business will be worth at the end of that period, and discount all of it back to today at a rate that reflects the risk. The total is the value of the enterprise. Subtract what is owed to lenders and you have the value of the equity. Divide by the number of shares and you have a value per share to compare with the market price.

Every word of that paragraph hides a judgement, which is why two competent analysts can build DCFs on the same company and arrive at values thirty per cent apart without either being wrong. The method is best understood by building one, so this article does: a complete valuation of a fictional company, with every number shown, followed by the questions a managing director or an interviewer would ask about it.

Where a DCF sits among the valuation methods

Bankers rarely use a DCF on its own. A valuation is presented as a range from several methods, each answering a different question. Trading comparables ask what the stock market pays today for similar listed companies, expressed as multiples of earnings or cash flow. Precedent transactions ask what acquirers paid for similar businesses in past deals, which usually includes a premium for control. A leveraged buyout analysis asks what a financial buyer could afford to pay and still earn its target return. A DCF asks what the business is intrinsically worth given its own forecast cash flows, independent of what the market happens to be paying this year.

That independence is both the strength and the weakness. A DCF can value a company with no good comparables, and it forces the analyst to make every assumption explicit. It also depends entirely on those assumptions, several of which (the growth rate in perpetuity, the discount rate) cannot be observed, so its output is a range that must be defended rather than a number that can be looked up. In practice the DCF is the method that gets argued about in the meeting, which is a reason to build it carefully.

The company: Alpenmetrik AG

Alpenmetrik AG is a fictional Swiss manufacturer of precision measurement instruments for industrial customers, with CHF 400 million of revenue in the year just ended, operating margins in the mid-teens, modest debt and 50 million shares in issue. It exists only in this article and in the Academy's teaching material. The assumptions below are chosen to be plausible for a business of that kind and simple enough to follow by hand; they are not a recommendation about any real company.

Step one: forecast unlevered free cash flow

Unlevered free cash flow is the cash the operations generate for all providers of capital, before any interest is paid or debt repaid. It is "unlevered" because it ignores the financing structure, which is what lets us value the enterprise first and deal with the debt afterwards. The definition the Academy teaches, and the one most banks use, is:

Unlevered free cash flow = EBIT × (1 − tax rate) + depreciation and amortisation − capital expenditure − increase in net working capital.

Each term has a reason. EBIT after tax is the operating profit the business keeps. Depreciation is added back because it was deducted to reach EBIT but no cash left the company. Capital expenditure is subtracted because cash did leave, to buy the assets that depreciation is spreading over time. The working-capital increase is subtracted because a growing company has to fund inventory and receivables before it collects from customers. Applied to Alpenmetrik, year by year:

Two things to notice. Free cash flow is lower than EBIT after tax in every year because the company is reinvesting more than it depreciates and tying up cash in working capital: that is what growth costs. And the forecast has structure: growth fades, margins rise gently, and the gap between capital expenditure and depreciation would close in a longer forecast as growth slows. A forecast whose lines do not relate to each other is the first thing an experienced reader notices.

Step two: the discount rate

The discount rate converts future cash into today's value. For unlevered cash flows, which belong to both lenders and shareholders, the rate is the weighted average cost of capital: the blended return the company has to earn to satisfy both. It has two components.

Cost of equity is estimated with the capital asset pricing model: the risk-free rate plus a premium for equity risk, scaled by how much this company's returns move with the market (its beta). For a Swiss franc valuation we take a long-dated Swiss government bond yield of 1.5% as the risk-free rate, an equity risk premium of 5.5%, and a beta of 1.2, re-levered from a set of comparable industrial companies to Alpenmetrik's own capital structure. Cost of equity = 1.5% + 1.2 × 5.5% = 8.1%.

Cost of debt is what the company would pay to borrow today, after tax, because interest is deductible. We take 3.0% pre-tax, which at an 18% tax rate is 2.46% after tax.

Weights come from the target capital structure at market values, not book values: here 90% equity and 10% debt. WACC = 0.90 × 8.1% + 0.10 × 2.46% = 7.5%. We round to 8.0% for the base case, which is slightly conservative, and show 7% to 9% in the sensitivity table, because an honest DCF treats the discount rate as a range from the start.

In an interview, the trap is to produce the number without the derivation. The interviewer wants to hear where beta comes from, why market weights, and why debt is cheaper (it is paid first, it is contractual, and the interest is deductible).

Step three: terminal value

The explicit forecast stops at year five, but the company does not. The terminal value captures everything after that in one number. The commonest method assumes the final year's cash flow grows at a constant rate for ever and applies the perpetuity formula:

Terminal value at year 5 = Free cash flow in year 5 × (1 + g) ÷ (WACC − g)

With g at 2.0%: 67.2 × 1.02 ÷ (0.08 − 0.02) = CHF 1,142 million. That is the value, at the end of year five, of all the cash flows from year six onwards. It still has to be discounted five years to today: 1,142 ÷ 1.08⁵ = CHF 777 million.

The alternative is the exit multiple method: assume the business is sold at the end of year five on a multiple of its EBITDA, taken from today's comparable companies. Year-five EBITDA is EBIT plus depreciation, 91.0 + 21.4 = CHF 112.4 million. Our perpetuity-growth terminal value of CHF 1,142 million implies a multiple of 10.2 times that EBITDA. If comparable industrial instrument makers trade between nine and twelve times, the two methods agree and the terminal value is defensible. If the implied multiple were twenty, something in the growth or discount assumptions would be wrong. Always compute the implied multiple as a cross-check; it is the first question a reviewer asks.

Step four: from enterprise value to a value per share

Discounting each year's free cash flow at 8% gives present values of 42.2, 43.8, 44.9, 45.6 and 45.7, which sum to CHF 222 million. Adding the discounted terminal value of CHF 777 million gives an enterprise value of roughly CHF 1,000 million. That is the value of the operating business to everyone who has a claim on it.

Shareholders come after lenders, so to get to equity value we subtract net debt of CHF 120 million: CHF 880 million. In a real valuation other claims sit in the same bridge: preferred shares, minority interests, unfunded pension deficits, leases depending on the accounting standard, and the value of any investments or surplus assets that the operating cash flows do not capture. Dividing CHF 880 million by 50 million diluted shares gives a value of CHF 17.59 per share. If the shares trade at CHF 14, the DCF says the market is undervaluing the business on these assumptions; at CHF 22, the reverse. Either way, the next step is not to announce the number but to show how much it moves.

Step five: sensitivity, and what the table is telling you

One percentage point on the discount rate moves the value per share by about 17% in either direction. Half a point on the terminal growth rate moves it by about 8%. Nothing in the five-year forecast has anything like that effect: doubling the year-three capital expenditure would change the answer by a franc or so. This is the central lesson of the DCF, and the reason experienced bankers spend their time on the terminal assumptions and the discount rate rather than polishing the forecast. It is also why the output is presented as a range, usually alongside the comparables and precedents in a "football field" chart, so the reader sees where the methods agree and where they do not.

Rather than stopping with time value of money and cash flow, the curriculum taught me how investment banks would value a company, and the practical challenges of arriving at a fair purchase price during and M&A deal.

Where DCFs go wrong

• A terminal growth rate above the economy's. A company growing at 4% for ever in a 2% economy eventually becomes the economy. Keep g at or below long-run nominal growth, and lower for a mature or cyclical business.
• Growth without investment. Revenue grows but capital expenditure and working capital stay flat, so free cash flow balloons. The forecast lines must relate: growth costs cash.
• Margins that only go up. Forecasts rarely include a bad year. Real businesses have them. A reviewer will ask what happens to the value in a downturn, and a model with no downside case has no answer.
• A discount rate from nowhere. A WACC must be derivable: risk-free rate, premium, beta source, debt cost, weights. "We used 10%" is not a derivation.
• Mismatched currency and rates. Swiss franc cash flows discounted at a dollar cost of capital, or nominal cash flows at a real rate. Keep everything in one currency and one basis.
• Forgetting the bridge. Enterprise value is not what shareholders get. Net debt, pensions, leases, minorities and preferred claims all come out first.
• Precision theatre. A value per share quoted to the centime, from a model whose discount rate is a guess to the nearest half point, misleads the reader about what the model knows. Present the range.
• Timing. Cash arrives through the year, not on 31 December; a mid-year convention is the usual fix. A valuation date part-way through a financial year needs a stub period. Both are details, but both are asked about.

How to say it in an interview

"Walk me through a DCF" is the most common technical question in investment banking interviews, and the answer takes about sixty seconds if you have built one. Forecast unlevered free cash flow for five to ten years, from revenue through operating profit after tax, adding back depreciation and subtracting capital expenditure and working-capital investment. Estimate a discount rate as the weighted average cost of capital, with the cost of equity from the capital asset pricing model and the cost of debt after tax. Calculate a terminal value by growing the final year's cash flow in perpetuity or applying an exit multiple, and cross-check one against the other. Discount the cash flows and the terminal value to today to get enterprise value. Subtract net debt and other claims to get equity value, and divide by diluted shares. Then say that the terminal value is usually the majority of the total and that the answer is a range driven by the discount rate and the terminal assumptions, and offer the sensitivity table. The follow-up questions will be about whichever step you said least confidently, which is a good reason to have built one on a company of your own before the interview.

During the program I acquired many skills including like multiple valuation methodologies and M&A strategy, as well as practice using them, which undergraduates do not often encounter.

Questions people ask

Why use unlevered free cash flow rather than net income?

Net income is an accounting figure after interest and non-cash charges, so it depends on how the company is financed and on depreciation policy. Unlevered free cash flow is the cash the operations generate for all capital providers before financing decisions, which is what the enterprise is worth to any owner.

What discount rate should I use?

The weighted average cost of capital for a company with this business risk and a sensible capital structure. In practice that means estimating the cost of equity from a risk-free rate, an equity risk premium and a beta drawn from comparable companies, the after-tax cost of debt from current borrowing rates, and weighting the two by target market values.

Why does the terminal value dominate?

Because a going concern is expected to generate cash for decades, and the explicit forecast only covers the first few years. Everything after the forecast horizon is compressed into one number. That is a feature of the method, not a mistake, but it means the terminal growth rate or exit multiple must be defensible.

When is a DCF the wrong tool?

When cash flows cannot be forecast with any confidence (early-stage companies, businesses in distress, commodity producers at a price extreme) or when the asset's value does not come from cash generation at all. In those cases comparables, asset-based methods or option-based approaches do more work.