Cash Flow Forecasting:
The integrated three-statement model

Why we forecast

Module 03 ended at the present moment. You can now read a 10-K and form a defensible view of what the firm's economics actually are. But every meaningful corporate-finance question is forward-looking: what is this company worth? Should we invest in this project? Can the borrower repay this loan? Should we acquire this competitor? None of those questions can be answered by looking at historical statements alone — they all require projecting the firm's future cash flows.

That projection is what a three-statement model does. It takes a small set of forward-looking assumptions about how the business will evolve — how fast revenue will grow, how margins will move, how much capital will be reinvested — and translates them into a complete projected income statement, balance sheet, and cash flow statement. The model's outputs are forecast statements; its inputs are the analyst's view of the firm's trajectory. The structure is what makes the analyst's view testable: any reader can examine the assumptions, agree or disagree, and see how the conclusions change.

Four use cases drive almost all three-statement modeling work:

What a three-statement model isn't

A three-statement model is not a forecast in the meteorological sense — it's not trying to predict what will happen. It's a quantification of an analyst's view. The model says: "if these assumptions hold, here is the cash-flow profile that follows." Different views produce different models. A pessimistic analyst's model and an optimistic analyst's model of the same company can both be internally consistent and mathematically sound; the difference is in the inputs, not the structure.

This means the value of the model is two-fold. First, it produces a numerical answer to a question (the firm's value, the borrower's coverage ratio, the project's NPV). Second, and more importantly, it makes the analyst's reasoning auditable. Anyone who disagrees can change a single driver and see the consequences. The model is a structured argument, not a conclusion.

The remainder of this module walks through the construction of an integrated three-statement model — what drivers to choose, how the statements link, what the common pitfalls are, and how to extract the free-cash-flow series that becomes the input to valuation. The Excel toolkit accompanying this module is a working version of exactly this model; we'll reference its structure throughout.

The driver-based approach

The fundamental design principle of a three-statement model is that everything traces back to a small set of drivers. Revenue, margins, working capital days, and capex intensity are the inputs; everything else — line items, subtotals, ratios, derived metrics — is computed from them by formula. A well-built model has perhaps 15-25 driver inputs and produces hundreds of outputs.

Why work this way? Three reasons:

• Testability. If you change a driver, the model recomputes everything coherently. You can run a sensitivity ("what if revenue growth is 100bp lower?") in seconds, not hours.
• Coherence. Hardcoding a forecast number into one cell creates a model that can become internally inconsistent — net income changes, but cash flow doesn't reflect it. Driver-based models stay coherent automatically because every cell is a formula.
• Communicability. The drivers are the conversation. Reviewers, partners, and counterparties argue about whether 8% revenue growth is realistic, not whether row 47 in the income statement is correct. The driver layer is where intellectual debate happens; the rest of the model just executes.

Three tiers of drivers

Most three-statement models organize their drivers into three logical tiers:

The Excel toolkit accompanying this module follows exactly this structure — the Assumptions tab is organized in these three tiers, with all yellow input cells gathered in one place. The Income Statement, Balance Sheet, and Cash Flow Statement tabs reference the assumption cells via formulas. Change a single driver value and watch eight years of projected statements update simultaneously.

Top-down vs. bottom-up

For revenue projections specifically, two philosophical approaches compete: top-down and bottom-up.

Top-down projections start with a market or industry size and apply a market-share assumption. "The global enterprise software market is $400B and growing 8% annually; our company has 3% share, expected to grow to 4% over five years." This produces clean revenue numbers but can be detached from operational reality.

Bottom-up projections build revenue from operational drivers: number of customers × average revenue per customer, units × price, stores × revenue per store. This connects the model to operating decisions but can become overwhelming for diversified firms.

Most practitioners use a hybrid: top-down sanity checks on bottom-up builds, or bottom-up for the main revenue line with top-down validation. The toolkit uses simple growth-rate assumptions for tractability, but a real corporate or investment model often elaborates the revenue build into multiple sub-components. Module 06 (Multiples Valuation) returns to the question of how granular revenue forecasts should be in different valuation contexts.

Building the income statement projection

The income statement is where projections start. Revenue at the top drives almost everything below it — costs and expenses are typically modeled as percentages of revenue, depreciation tracks revenue indirectly through the capex roll-forward, and net income flows into the cash flow statement and retained earnings. Build the income statement first, then the cash flow statement, then the balance sheet — that's the standard order and it's standard for a reason.

The structural pattern

The mechanics work top-to-bottom in the same sequence Module 03 used to read existing statements:

That is, in essence, the entire income-statement projection. The conceptual challenge lies far more in the assumptions than in the arithmetic itself. Two analysts building this same model will produce very different projections because they choose different growth rates, different margins, and different tax assumptions. Picking those numbers well requires understanding the business, not understanding accounting.

The toolkit's Income Statement tab implements this pattern across eight years. Click any cell with black text to see the formula — you'll find that revenue in column G is just =F5*(1+Assumptions!G$8), EBIT is =G9+G10, and net income is =G13+G14. There's no magic. The model executes the pattern eight times across columns B-I.

How to choose growth rates

Revenue growth is the most consequential driver in any model — small differences compound dramatically over five years — and it's the driver analysts spend the most time on. A few principles:

• Anchor to recent history. If the firm grew 8% last year and 7% the year before, projecting 15% next year requires explanation. What's changing? New product launch? Market expansion? Recent investments paying off? Without a story, regress toward recent reality.
• Reflect maturation. Even fast-growing firms slow down. A 30% grower today won't be growing 30% in five years — the law of large numbers makes that mathematically nearly impossible. The toolkit's default growth path (10%, 9%, 8%, 8%, 7%, 6%, 5%) reflects normal maturation.
• Don't exceed industry growth indefinitely. In the long run, a firm cannot grow faster than its industry without monopolizing it. If your projection has the firm at 10× industry growth in year 5, the projection is essentially a story about industry transformation, not company performance.
• Be wary of hockey-stick projections. Models that show flat or declining revenue followed by a sudden inflection upward are usually wrong. Real firms don't suddenly start growing faster. The hockey-stick pattern is a tell that the analyst is forcing the numbers.

How to project margins

Margin assumptions follow similar logic: anchor to history, reflect dynamics. The key questions:

• Is the gross margin stable, expanding, or compressing? Companies with pricing power tend to maintain or expand gross margins. Commoditizing businesses see them compress. Look at the trailing 5-year history first.
• Is operating leverage real? Many growth companies argue that SG&A as % of revenue will decline as revenue scales (operating leverage). This is sometimes true (software, marketplaces) and sometimes false (services, retail). Look at whether SG&A grew slower than revenue historically.
• What's the steady-state tax rate? Most US firms run effective tax rates in the 21-28% range. Some have unusually low rates due to international structuring (older Apple, Google) or high rates due to specific geographies. The historical effective rate is usually a reasonable starting point.

The toolkit's defaults reflect a maturing consumer/industrial firm: gross margin gradually expanding from 30% to 32% over 8 years, SG&A declining from 18% to 16% (modest operating leverage), tax rate flat at 25%. These are starting points — your task is to replace them with assumptions calibrated to your specific company.

The result: a projected income statement

Here's what comes out the other end, using the toolkit's default sample assumptions:

Eight years of net income, generated from a small set of drivers. This output now becomes input to the rest of the model. Net income flows to retained earnings on the balance sheet (Module 03's bridge) and to the starting line of the cash flow statement. The income statement is upstream of everything else.

Building the balance sheet projection

The balance sheet is the trickier statement to project. Its line items don't all behave the same way: some scale with revenue (working capital), some roll forward from period to period (PP&E, retained earnings, cash), and some are held constant in simple models (debt, common stock). Cash is the line item that integrates the cash flow statement back into the balance sheet — it equals prior-year cash plus the net change in cash from the cash flow statement, by construction.

Three projection patterns cover most balance sheet items:

Pattern 1 — Working capital (scale with operations)

Receivables, inventory, and payables scale with operating volume. The standard analyst approach uses days metrics:

The implicit assumption: working-capital intensity is stable. If the firm collects from customers in 45 days today, it collects in 45 days next year — unless you change the DSO assumption. This is usually a defensible starting point but breaks down in special cases (retailer cyclical inventory build, SaaS firm with deferred revenue, contractor with progress billing).

The toolkit's default working-capital path has DSO at 45 days throughout, DIO declining from 60 to 55 days (operational improvement), and DPO expanding from 40 to 45 days (better supplier terms). Working capital becomes incrementally more efficient over the projection — a typical management commentary in real annual reports.

Pattern 2 — Long-term assets (roll-forward)

Property, plant, and equipment rolls forward by adding capex and subtracting depreciation:

This is the model's most important balance-sheet equation because it links capex (a cash flow item) to PP&E (a balance sheet item) to D&A (an income statement item). The roll-forward enforces consistency: capex spent today becomes PP&E that gets depreciated over future periods, with D&A flowing back to the income statement.

Long-run, capex and D&A converge — D&A in any given year reflects depreciation of all prior-year capex, and in steady state, capex ≈ D&A. A model where capex permanently exceeds D&A implies the firm is growing its asset base (capacity expansion); a model where capex is permanently below D&A implies the firm is shrinking. Neither is wrong, but each tells a different story about the business.

Pattern 3 — Equity (roll-forward via net income and dividends)

Retained earnings is the bridge from the income statement to the balance sheet:

Common stock and paid-in capital, in simple models, are held constant — no equity issuance or buybacks during the projection. This is the toolkit's default. More elaborate models include explicit issuance and repurchase schedules; for now, treat constant equity as a working assumption.

Debt — the simple version

The toolkit holds long-term debt constant at its opening balance. This is the simplest possible debt assumption: no new issuance, no repayment, no refinancing. Real corporate models often include a detailed debt schedule with principal amortization, new issuance to fund capex or shortfalls, and revolving credit facilities for working capital. For valuation purposes, the simplification rarely matters much — the present value of capital-structure decisions tends to be a second-order effect compared to operating performance.

Cash: how the three statements close the loop

Here's the subtle point that trips up many first-time modelers. The three statements together form a closed accounting system. Once you have the income statement and you project every balance-sheet line item except cash — working capital from days, PP&E from roll-forward, debt held constant, common stock held constant, retained earnings via net income and dividends — the cash flow statement is fully determined: it reconciles net income to the period-over-period change in every other balance-sheet line. Cash on the balance sheet then follows from the cash flow statement.

So cash on the balance sheet is not a free variable. It is a roll-forward: prior-year cash plus the net change in cash computed by the cash flow statement. If the firm generates more cash than expected (lower working-capital absorption, stronger operating performance, less capex), the cash flow statement shows it and cash on the balance sheet grows accordingly; if it generates less, cash shrinks. The cash flow statement tells you the change in cash; the balance sheet records the result.

This is how the three statements integrate in the toolkit. The cash flow statement reconciles exactly to the balance sheet's cash line — by construction, not by coincidence.

The result: a projected balance sheet

Continuing with the toolkit's sample assumptions, here's what the projected balance sheet looks like:

Total assets equals total liabilities plus equity in every single year — the equation balances by construction. The toolkit's Balance Sheet tab includes a "balance check" row at the bottom that displays the difference; if your assumptions produce a non-zero balance check, something has gone wrong with the linkages.

Building the cash flow statement

The cash flow statement is what makes a three-statement model an integrated model rather than three loosely-connected projections. Build it last, because every line item references either the income statement or two consecutive periods of the balance sheet.

Operating cash flow — the indirect method

Real-world cash flow statements use the indirect method: start with net income, adjust back to cash. Three categories of adjustment:

Each working-capital adjustment captures the difference between accrual and cash. If receivables grew $20M during the period, that means $20M of revenue was recognized but not yet collected — subtract from cash. If payables grew $15M, that's $15M of expense recognized but not yet paid — add to cash. The cash flow statement is the bridge from accrual back to reality.

The toolkit's Cash Flow Statement tab implements this exactly. Net income comes from the Income Statement tab. D&A is added back. Working-capital changes are computed as differences between consecutive years on the Balance Sheet. The resulting OCF line is what valuation analysts consult most.

Investing cash flow — capex matters

For most non-financial firms, investing cash flow is dominated by capex. The toolkit projects capex as a percentage of revenue and shows it as the only investing line. More elaborate models include acquisitions, asset sales, and investment-security purchases, but for valuation purposes, capex is what matters.

Capex's relationship to D&A is what reveals the firm's growth profile:

• Capex > D&A means the firm is growing its asset base. Net PP&E expands. The firm is investing for future capacity. Most companies during growth phases.
• Capex ≈ D&A means the firm is in maintenance mode — replacing assets as they age, not expanding. Steady-state mature firms.
• Capex < D&A means the firm is harvesting — using up its asset base without replacing it. Either declining business or temporary discipline. Persistent under-investment is a warning sign.

The toolkit's defaults run capex slightly above D&A (capex 5-6% of revenue, D&A 4%), reflecting modest capacity growth. PP&E grows from $400M to $525M over 8 years — consistent with a maturing firm continuing to invest.

Financing cash flow — keep it simple

The toolkit's financing section has only one line: dividends paid. Real-world financing sections include debt issuance, debt repayment, equity issuance, share buybacks, and dividend payments. For most operating projections, the simplification works because most operating decisions don't require financing changes. When you do need to model financing more elaborately — leveraged buyouts, recapitalizations, accretive M&A — you add complexity selectively.

The reconciliation that makes it a model

Operating cash flow + Investing cash flow + Financing cash flow = Net change in cash. The net change flows to the balance sheet: Year (t) cash equals Year (t-1) cash plus the period's net change. This single connection is what integrates the three statements.

The toolkit passes this test. Open it, change the Year 5 gross margin from 32% to 33% on the Assumptions tab, and watch eight years of statements recompute with the balance sheet still balancing exactly. That's what integration means.

Free cash flow: FCFF vs. FCFE

The point of all this model-building is to produce a free cash flow series — the cash genuinely available for distribution to capital providers. Module 03 introduced free cash flow at a high level; this section sharpens the distinction between two flavors that valuation work depends on.

FCFF — Free Cash Flow to the Firm

FCFF is the cash flow available to all capital providers — both debt holders and equity holders — before any distributions. It treats the firm as a single economic entity, abstracting away from how it's financed.

NOPAT — Net Operating Profit After Tax — is the right starting point because it captures operating profitability after taxes but before the impact of debt. Adding back D&A removes the non-cash expense; subtracting capex captures real reinvestment; subtracting the change in working capital captures the cash tied up in operations.

FCFF is the standard input to enterprise-value DCF (Module 07). When you discount FCFF at the firm's cost of capital (WACC, Module 05), you get enterprise value — the value of the entire business operation, debt and equity combined.

FCFE — Free Cash Flow to Equity

FCFE is the cash flow available specifically to equity holders, after debt holders have been paid. It accounts for interest payments, debt principal repayments, and any new debt raised.

FCFE is the input to equity DCF, where you discount at the cost of equity directly (rather than WACC) and arrive at equity value. Both FCFF and FCFE produce the same valuation when applied correctly — they're two paths to the same answer. FCFF/WACC is the standard for most public-company valuations because it's less sensitive to capital-structure assumptions; FCFE/cost-of-equity is more common in financial-services contexts where the capital structure is integral to the business model.

Which to use, when

The toolkit's FCF Summary tab computes both FCFF (used in Module 07) and the simpler "FCF = OCF − Capex" version (a quick estimate). Look at the toolkit's Year 5 numbers: FCF = $80M, FCFF = $88M. The small difference reflects working-capital adjustments and the exact treatment of D&A. Both definitions tell the same story — but which precise number you use depends on what valuation method you're applying.

The toolkit's projected FCFF series — $50M in Year 1 growing to $125M in Year 8, with cumulative FCFF of $501M across the 5-year projected period — is exactly the kind of input that flows into Module 07's DCF valuation. Save the toolkit; you'll be using it again.

Common pitfalls & professional discipline

Most three-statement models built by junior analysts have the same handful of problems. The math works, the formulas execute, the balance sheet balances — but the economic content of the projection is unrealistic. Six pitfalls account for most of the trouble:

Professional discipline: the practitioner's mindset

Beyond avoiding specific errors, four habits separate experienced modelers from novices:

• Always sanity-check against the past. Before believing any projection, look at the historical 3-5 years. Are projected growth, margins, and capital intensity in the same neighborhood as history? If not, why? "Why" should be a story, not a hope.
• Run sensitivity analysis as standard practice. Never present a single forecast number; always show how the answer flexes with the most consequential drivers. Revenue growth ±2pp, gross margin ±100bp, terminal growth rate ±50bp. The toolkit's Sensitivity tab is a starting template.
• Keep the model auditable. Driver inputs in one place (Assumptions tab). Color-coded inputs and formulas so reviewers can distinguish them at a glance. No magic numbers buried inside formulas. The model should be readable by someone other than you.
• Be ready to defend every assumption. If a partner asks "why 8% revenue growth?", you should have an answer that goes beyond "that's what management guided to". The driver layer is where intellectual debate happens; arrive prepared.

Professional discipline is what makes the difference between a model that produces a number and a model that informs a decision. The number is the output; the decision is the point.

Six firms, six forecasting challenges

The driver-based approach is universal, but the right driver structure varies by industry and country. Here's how six well-known firms challenge the standard template:

What unites these six cases: the structure of the three-statement model is identical, but the drivers require company-specific design. There's no universal "growth rate" assumption that fits every firm. Building a useful model means knowing the business well enough to choose the right drivers, then executing the standard mechanics on top of that driver structure.

The bridge to Module 05

You now have a working three-statement model and a free-cash-flow projection. What's missing for valuation? The discount rate. A $100M FCF five years from now is worth less than $100M today — but how much less? The answer depends on the firm's cost of capital, which depends on its capital structure, the riskiness of its business, and the prevailing risk-free rate. Module 05 builds out cost of capital — the WACC and CAPM frameworks that turn the FCF projection into a present value.