Discounted cash flow analysis: five steps for stock valuation

It happens earlier: analysts discount a firm-level cash flow at the cost of equity, treat EBITDA as cash flow, or forget that debt holders have a prior claim on enterprise value.

The spreadsheet still calculates. The valuation is still wrong.

DCF is not a device for discovering an objectively correct stock price. It is a disciplined way to state what a business must deliver in cash, how risky those cash flows are, and what remains for shareholders after other claims are settled. Its output is conditional. The conditions matter more than the final cell.

The practical workflow has five steps:

1. Select the cash flow claim.

2. Forecast the explicit period.

3. Match the discount rate to that claim.

4. Calculate terminal value without smuggling in impossible assumptions.

5. Reconcile enterprise value to equity value per share.

1. Define the cash flow foundation: FCFF versus FCFE

The first question is basic and routinely mishandled: whose cash flow is being valued?

A company has operating cash flow. Creditors have contractual claims. Preferred holders may have claims. Common shareholders receive the residual. These are not interchangeable layers.

A firm-level model uses free cash flow to the firm, or FCFF:

FCFF = EBIT × (1 − tax rate) − (capital expenditures − depreciation) − change in non-cash working capital

FCFF measures cash generated by operations after taxes and reinvestment, before interest payments and net debt financing. It belongs to all capital providers. The result is enterprise value.

An equity-level model uses free cash flow to equity, or FCFE. It starts from cash available to common shareholders after interest, debt repayments, and net borrowing. The result is equity value directly.

The distinction is not academic. It determines the entire architecture of the model.

FCFF is usually cleaner for businesses with changing capital structures, significant debt, or a history of refinancing. It isolates operating performance from the financing decisions layered on top of it. That is useful when a company’s interest expense has been distorted by a recent acquisition, a sharp rate reset, or temporary balance-sheet stress.

FCFE can work well for a stable business with predictable leverage. But it requires more attention than it receives. A company that issues debt to fund share repurchases can show higher FCFE without creating additional operating value. The cash reached shareholders because leverage changed, not because the business improved.

A second distinction matters inside FCFF: EBIT is not EBITDA.

EBITDA excludes depreciation and amortization. That is tolerable for a narrow operating comparison. It is inadequate as a valuation cash flow. Asset-heavy businesses must replace worn equipment. Software businesses may capitalize internal development costs. Retailers may consume cash through inventory growth and supplier terms. Those reinvestment requirements do not disappear because EBITDA is high.

EBITDA can describe operating scale. It cannot settle the question of how much cash survives reinvestment.

The forensic work begins with the historical statements. Before forecasting anything, identify whether reported earnings have been flattered by low capital expenditure, released working capital, capitalized costs, or non-recurring tax items.

A useful starting review includes:

• Depreciation relative to capital expenditure. Persistently low capex may be valid for an asset-light business. It may also mean maintenance investment has been deferred.
• Working capital behavior. A cash conversion cycle that improves because payables stretch faster than inventory and receivables is not a permanent source of cash.
• Capitalized costs. Capitalized software, commissions, content, or development expenditure may lift current EBIT and defer expense recognition into later periods.
• Impairments and restructuring charges. An impairment is non-cash when recorded, but it often confirms that past capital allocation was cash-destructive.
• Stock-based compensation and dilution. A non-cash expense can still transfer a material claim from existing shareholders to employees.

The model should not repair a weak business. It should expose it.

2. Build explicit forecasts for the three- to five-year horizon

A normal DCF uses an explicit forecast period of roughly three to five years. Beyond that point, detailed annual estimates tend to become a performance of confidence rather than analysis.

The explicit period is where the model earns its existence. Terminal value will later compress the distant future into one formula. The near-term forecast must therefore carry the operational evidence: revenue growth, margins, taxes, reinvestment, and working capital.

The proper sequence is simple.

Start with revenue, but do not end there

Revenue is the visible line. It is also the line most often extrapolated mechanically.

A forecast should distinguish between volume, price, mix, acquisitions, and currency where those drivers are material. A company that has grown revenue through acquisitions does not have the same organic earnings power as one that has expanded unit volumes. A company reporting price-led growth may be protecting margin today while weakening demand tomorrow.

There is no universal correct revenue-growth assumption. The point is to make the assumption traceable.

For a mature firm, the base case should explain why growth persists rather than merely extend the last reported quarter. For a cyclical firm, normalized revenue may matter more than current revenue. For a high-growth firm, the core issue is usually not the next year’s growth rate. It is the amount of incremental capital required to sustain that growth.

Forecast operating margins separately from sales

Revenue does not become FCFF by passing through a margin assumption. The margin needs its own logic.

A rising EBIT margin can come from scale, pricing, lower input costs, reduced sales spending, or accounting reclassification. These are different mechanisms with different durability.

The model should ask:

1. Is gross margin changing because of product mix or temporary input costs?

2. Are selling and administrative costs genuinely scaling, or have expenses been deferred?

3. Are research and development costs expensed, capitalized, or shifted between categories?

4. Does the stated tax rate reflect recurring cash taxes, or a temporary benefit?

5. Does the business require rising investment to protect its margin?

This is where management-adjusted earnings should be treated with suspicion. “Adjusted” often means the cash cost has been moved outside the preferred narrative, not outside the business.

Forecast reinvestment, not just earnings

A DCF fails when it assumes revenue expansion without funding the assets required to produce it.

FCFF deducts net capital expenditure and changes in non-cash working capital for a reason. Growth consumes cash unless the business has an unusual negative working-capital model or a demonstrably asset-light structure.

For each projected year, forecast:

• EBIT and cash taxes;
• depreciation and amortization;
• capital expenditure;
• inventory, receivables, and payables behavior;
• other operating assets and liabilities where material.

The direction matters as much as the level. A company can report strong operating cash flow for one year because receivables were collected faster or inventory was liquidated. That does not establish a recurring free-cash-flow margin.

A simple forecast table forces the logic into view:

The forecast should be internally consistent. Higher growth with lower capex, better margins, stable working capital, and no dilution is possible. It is not the default condition. It needs evidence.

3. Match the discount rate to the cash flow model

The discount rate is not a generic “risk percentage.” It is the required return for a specific claim on a specific stream of cash flows.

For FCFF, the correct rate is the weighted average cost of capital:

WACC = Cost of Equity × E / (D + E) + After-tax Cost of Debt × D / (D + E)

Here, E and D are market values of equity and debt. The debt component is after tax because interest is generally tax-deductible at the corporate level.

For FCFE, the correct rate is the cost of equity.

This matching rule is non-negotiable:

Discounting FCFF at the cost of equity usually understates present value because the cost of equity is normally higher than WACC. Discounting FCFE at WACC does the reverse. The error can be large enough to dominate the apparent insight in the rest of the model.

The WACC itself should not be copied from a market data terminal and accepted as settled fact. It is an assumption built from assumptions.

The cost of equity depends on the risk-free rate, equity risk premium, beta or another risk measure, country exposure where relevant, and company-specific risk judgment. The cost of debt depends on actual borrowing costs, credit risk, maturity structure, and the tax treatment of interest. Capital weights depend on market values, not stale book values in a balance sheet footnote.

A frequent shortcut is to use a company’s current debt-to-equity ratio forever. That is weak practice for firms undergoing deleveraging, acquisition-led expansion, or recurring share repurchases. The discount rate should reflect a sustainable capital structure, not necessarily the balance sheet at one reporting date.

There is no universal WACC. There is no correct rate that applies across sectors, currencies, capital structures, and profitability profiles. A low-debt utility, a profitable software company, and a leveraged retailer do not deserve the same discount rate because a template says 8%.

A DCF is not conservative because the WACC looks high. It is conservative only when the cash-flow assumptions and the discount rate describe the same business.

The practical test is blunt: if the model uses optimistic operating assumptions but a punitive discount rate, it is not balanced. It is incoherent. A high WACC does not compensate for invented cash flow.

4. Calculate terminal value without hiding the valuation inside it

Terminal value is usually the largest component of a DCF. This is inconvenient, because it means the model’s conclusion often depends more on the steady-state assumption than on the detailed forecast.

Under the perpetual-growth method:

Terminal Value = FCF in terminal year × (1 + g) / (discount rate − g)

For an FCFF model, the discount rate is WACC.

The formula is mechanically simple. The economic discipline is not.

The stable growth rate must remain below the discount rate. More fundamentally, it cannot exceed the long-run growth capacity of the economy in which the business operates. A mature company cannot compound free cash flow faster than its economic environment forever while retaining stable competitive conditions and a stable capital base.

The denominator explains the sensitivity. As g approaches the discount rate, terminal value expands sharply. A small adjustment can produce a large valuation swing. This is mathematics, not discovery.

A proper terminal-period assumption includes more than a growth rate. It needs a stable operating profile:

• revenue growth converges toward a sustainable rate;
• EBIT margins settle at a plausible mature level;
• cash taxes normalize;
• reinvestment supports the assumed growth;
• working capital no longer generates one-off releases;
• capital structure is stable enough for the selected WACC.

The reinvestment point is regularly omitted. If a company is assumed to grow perpetually, it must reinvest perpetually. A terminal model that combines durable growth with negligible capex and no working-capital need is often just an inflated annuity.

An exit-multiple method can also estimate terminal value. For example, an analyst may apply an EV/EBITDA multiple to the final forecast year. That is useful as a market-based cross-check. It is not a purely intrinsic DCF input.

Using an exit multiple imports relative valuation into the model. The resulting value depends on what comparable firms trade for at that time, including any broad market overvaluation or compression. There is nothing improper about this. It should simply be labeled correctly.

Sector EV/EBITDA ranges vary widely. There is no single “normal” multiple. The difference is driven by capital intensity, margin structure, growth, leverage, cyclicality, and reinvestment needs. Applying a fashionable multiple to an unrelated company does not create comparability.

A sound model runs both a perpetual-growth terminal value and a market-multiple cross-check. If the outputs diverge sharply, do not average them automatically. Find the source of the disagreement. It may be the forecast margin, the reinvestment assumption, the chosen multiple, or the market’s current pricing.

5. Reconcile firm value to equity value per share

A FCFF DCF produces enterprise value. That is the value of the operating business before allocating claims among capital providers.

It is not yet the value of common equity.

The reconciliation begins with enterprise value and removes non-equity claims. Net debt is the obvious item, but it is not always the only one. Depending on the company, the analyst may need to account for preferred equity, non-controlling interests, pension deficits, lease obligations where consistently treated, and other senior claims.

Cash and marketable securities may need to be added because operating FCFF does not include non-operating cash. But not every cash balance deserves a full addition. Some cash is operational. Some may be trapped in foreign subsidiaries. Some may be needed to cover liabilities that have not been properly adjusted elsewhere.

The bridge should be explicit:

1. Add the present value of forecast FCFF.

2. Add the present value of terminal value.

3. This produces enterprise value.

4. Add non-operating cash and marketable securities where appropriate.

5. Subtract debt and other non-equity claims.

6. Divide the resulting equity value by diluted shares outstanding.

The last line can still mislead if the share count is careless.

Basic shares are rarely the right denominator for a company that pays heavily in equity or has a large option overhang. The model should use a diluted share count that reflects instruments likely to become claims on equity value. A buyback program should not be projected as automatic value creation if it is funded by debt or conducted at prices above intrinsic value.

This reconciliation is where many polished DCFs quietly lose discipline. Enterprise value is presented beside the market capitalization. Net debt is ignored. Cash is added twice. Lease liabilities appear in WACC but not in the valuation bridge. The arithmetic may be neat. The claim structure is not.

The intrinsic value is a range, not a verdict

The final output of discounted cash flow analysis should be a valuation range built from sensitivity analysis, not a theatrical target price.

At minimum, vary WACC and terminal growth. For businesses with unstable economics, vary the operating margin and reinvestment rate as well. The purpose is not to create a wide grid that gives the appearance of rigor. It is to identify which assumption actually controls the result.

If a one-point change in WACC or a modest change in terminal growth reverses the investment conclusion, the stock does not have a robust margin of safety. It has a fragile spreadsheet.

The sober conclusion is usually narrower than the model suggests. A DCF can establish that a business is priced for unusually high margins, low reinvestment, or an implausibly durable growth runway. It can also show that the market price already discounts a harsh operating outcome. Those are useful findings.

But the intrinsic value estimate remains conditional: forecast FCFF, matched discount rate, sustainable terminal assumptions, net claims, and diluted shares. Remove any one of those disciplines and the DCF becomes a valuation multiple with extra formatting.