EI 8 Financial Statement Forecasting in Equity Valuation

Earlier equity modules valued a firm in one of two shortcut ways. A present value model took a base year cash flow, such as a dividend or free cash flow, and grew it at one or more assumed rates over a horizon. A relative value model multiplied a base year metric by a market multiple. Both are quick, but each rests on the assumption that the relationships among the underlying value drivers stay fixed.

This module takes a different route. Instead of assuming one aggregate growth rate, the analyst forecasts the whole set of financial statements into the future and lets cash flow emerge from the pieces. These are called disaggregated valuation models, because operating, investing, and financing performance are each projected separately, and overall growth is the combined result rather than a single input. The payoff is a valuation that carries an explicit narrative: the analyst can say exactly where margin gains, heavier investment, or a change in financing are expected to come from.

From forecast statements to a value

A disaggregated model builds each future free cash flow to the firm (FCFF) from income statement and balance sheet components, then feeds those cash flows into a familiar valuation engine. The projected cash flows can serve as the numerator of a discounted cash flow estimate of intrinsic value:

Alternatively, a forecast income measure can sit in a multiple used for relative value:

Because a disaggregated model usually produces revenue and EBITDA or EBIT forecasts anyway, an analyst can combine a projected income figure with a peer multiple to set the terminal value, a step we return to later.

The rationale: different pieces, different growth

The central argument for the approach is simple. Whenever the components of cash flow are expected to grow at different rates, the overall growth rate cannot be captured by a single number, and a disaggregated model will give a different, more informative answer than a constant growth model. Consider consensus estimates for Amazon.com in May 2024: the Q1 2025 revenue forecast of USD158.989 billion was about 10.9 percent above actual Q1 2024 revenue of USD143.313 billion, while the operating profit forecast of USD18.274 billion was roughly 19.4 percent above the year-earlier USD15.307 billion. Revenue and operating profit are pulling at visibly different speeds, exactly the situation a disaggregated model is built for.

A disaggregated view also captures changes in policy, not just in operating growth. Take a mature firm with a stable payout ratio whose earnings growth is expected to fade. With fewer attractive projects to fund, the analyst may reasonably assume the firm lifts its dividend payout ratio, so dividends per share grow faster than earnings for a stretch. The next example works that idea through.

Building the model: the areas of focus

Model construction begins with the analyst choosing how much detail to carry, area by area. A disaggregated FCFF forecast usually rests on four blocks: revenue, expenses or profit margin, asset growth and efficiency, and financing. The first two produce the income assumptions behind the cash flow metrics; the third adds investment; the fourth specifies the debt and equity used to fund the plan. At its simplest, FCFF combines net operating profit after taxes (NOPAT) with net investment.

More detail is a double-edged tool. Extra line items let the analyst tell a richer story and pull more levers, but they also add noise, reduce how easily the output can be read, and demand more frequent updating as relationships among variables shift. Complexity should be added only where it earns its keep. The following example shows how even a modest simplification can distort a forecast if applied without thought.

Revenue is the first line to project and the one drawing the most scrutiny, because it drives every downstream cash flow measure. Three broad approaches are common: a historical approach, a bottom-up approach, and a top-down approach.

Historical approach

For mature companies, those with established products, a settled product mix and geography, and no expected shift in the competitive landscape, past revenue is a reasonable base for future estimates. Nestle, the world largest food company, is a good case: it earned just over CHF92 billion in 2012 and about CHF93 billion in 2023, a mean annual growth rate of 0.16 percent, a compounded rate (CAGR) of just 0.08 percent, and a 4.22 percent standard deviation of yearly growth. With revenue that scaled and that stable across roughly 190 jurisdictions, history is a defensible starting point. Even so, historical growth should be tempered with the company strategy, its industry, and the outlook for its products, so that both opportunities and threats reach the forecast.

History also applies segment by segment. Firms report separate financials for material operating segments, and those segments can face very different prospects. At Amazon.com, the Online Stores and Physical Stores segments are expected to grow slowest, while Amazon Web Services and Advertising Services drive double-digit percentage growth off a base of nearly USD575 billion in 2023 revenue. Blending one growth rate across such different lines would misstate the whole.

Bottom-up approach

Sometimes revenue is better assembled from fundamental units of activity. Airline analysts forecast available seat kilometers (ASKs), the number of seats offered multiplied by the kilometers those seats are flown over a period, then multiply an estimate of revenue per ASK by forecast ASKs to reach revenue. Capacity-based, bottom-up forecasts also suit businesses tied to physical locations. Retail and restaurant analysts watch same-store sales growth closely: when Starbucks reported weak same-store sales in April 2024, its shares fell 12 percent, and over the long run its revenue growth tracks same-store sales more than new store openings.

Top-down approach

A top-down forecast starts from the size of the whole market, often within a macroeconomic view, breaks it into units and price per unit, and then applies the company market share by segment. The Toyota example below shows the mechanics.

Expenses and the profit margins they produce are the next key drivers. Constant growth models force expenses to grow with revenue; a disaggregated model lets gross and operating margins rise or fall relative to the base year. Three routes are common: a historical approach, itemizing the economically significant expenses, and assuming margins directly.

Historical and regression approaches

Growing each cost by its own past rate ignores the link between revenue and expenses. A better historical method regresses the change in a cost line on the change in revenue.

Itemizing significant expenses

Standardized statements from a data vendor can bury the costs that matter most. Lufthansa is again instructive. A standardized 2023 income statement shows COGS, D&A, and SG&A but no separate lines for the two expenses that dominate an airline: labor and jet fuel. Switching to the as-reported statement surfaces staff costs of EUR8,344 million, yet fuel is still hidden inside cost of materials and services of EUR20,378 million. Only by going to the company own annual report does the analyst find jet fuel costs of EUR7,931 million. Because fuel is largely unrelated to revenue while labor moves more closely with it, modeling them separately matters, and it can only be done once they are pulled out of the aggregates.

Direct margin assumptions

Rather than forecasting each expense, an analyst can assume the margin directly, since EBIT margin drives both FCFF and FCFE valuations. A direct margin assumption still implies something about the underlying expense growth, as the next example shows. Note too that net margin brings in items outside operations, such as write-downs, restructuring or legal costs, interest, and taxes, so net margins can move for reasons unrelated to the core business.

Direct margin work becomes more powerful once cost behavior is split into fixed and variable parts, because the same revenue swing then produces very different margins. The next example refines the Jelenico case with a cost structure and three revenue scenarios.

Income assumptions are only half the model. The balance sheet supplies the investment and financing that turn profit into free cash flow. The asset side estimates how much working capital and long-term assets a firm needs; the financing side specifies how any shortfall or surplus is covered.

Asset models

Assets can be projected by regression too, mirroring the expense equation.

Some firms must grow assets faster than revenue to hit their plans; others invest proportionately less. A quick test compares the CAGR of key asset measures against the CAGR of revenue: a firm whose assets have historically outgrown revenue is more investment-hungry than a peer with the opposite pattern. Asset regressions are trickier than expense regressions, because a purchase of capacity and the revenue it later generates may fall in different years. For Deutsche Lufthansa, a regression of gross PP&E change on sales change gives an intercept of 0.0478 and a slope of only 0.1716 that is not statistically significant, implying most PP&E growth is unrelated to that year revenue. Applying it produces a deliberately smoothed, stable investment path: assumed sales growth of −5, −1, 3, 7, and 11 percent maps to gross PP&E growth of 3.9, 4.6, 5.3, 6.0, and 6.7 percent. The takeaway is that Lufthansa spends on aircraft far more steadily than its cyclical revenue swings.

An itemized asset model asks two questions of each balance sheet line: does it arise from the operating process, and does it involve cash payments or receipts? Where both answers are yes, the item is projected to grow with the business. Non-operating assets such as marketable securities are generally not grown, except to absorb changes in cash under the financing model. Items that move on market value rather than cash, such as goodwill after an impairment, are not modeled as investment.

Financing models

Financing assumptions are essential for FCFE, which can be written two equivalent ways.

A simple way to size a year financing need is to forecast the change in operating assets, subtract the change in non-debt liabilities, and subtract the change in retained earnings (net income less dividends).

A positive need is usually met from cash or debt rather than equity, since firms issue equity infrequently because of cost, dilution, and timing. An analyst must still guard against running cash too low or debt ratios too high. A surplus (negative need) is typically returned through stock repurchases or, sometimes, higher dividends. The table below summarizes the choices.

These moves ripple differently through the two cash flow metrics. Because FCFF depends only on profitability and investment, it is largely insensitive to how a financing need is met; it changes only slightly through financing costs, which these simplified examples ignore. FCFE, by contrast, responds directly to debt, dividend, and equity assumptions.

The three building blocks of a valuation, the cash flow numerator, the discount rate denominator, and the terminal value, are adjusted to fit the subject company. The single biggest driver of those adjustments is where the firm sits in its life cycle, because the available data, the relevant metric, and the value driver all change from stage to stage.

The standard valuation approaches generally assume a mature firm. The other stages need adjustment.

Start-up phase

A start-up often has little revenue, negative cash flow, no operating record, and a high chance of failure, so models built on existing statements are of little use. Value comes from future potential, so the analyst usually skips a near-term cash flow forecast and anchors on a future terminal value built from a price-to-sales multiple.

Returns with no interim cash flows can be stated as an annualized internal rate of return over n periods, and the target multiple of money, or return on investment (ROI), links directly to it.

Because these ventures are risky, private and venture capital (VC) investors often demand an ROI above 10 times over five years or more. When a new investor buys in, the analyst also distinguishes value before the money from value after it.

Growth phase

Growth firms have a proven product and real revenue, and they raise equity to expand assets and reach sustainable profitability. Unlike start-ups, they can lean on financial statements, and they target a terminal value from an earnings multiple rather than a sales multiple. Differing growth rates across statement lines give a richer picture than a simple two-stage model.

Decline phase and the probability of distress

Firms in decline face falling sales, lost share, thinner margins, and a real risk of financial distress, meaning the firm can no longer cover what it owes its debtholders. Distress may end in sale, restructuring, or liquidation, and if asset proceeds fall short of debt, shareholders can lose everything. Standard models struggle here, so three areas need care: cash flow adjustments for impaired or lost assets and customers; a denominator that is hard to pin down as default risk swings both the cost of debt and the required return on equity; and terminal values, since going-concern assumptions and peer multiples distort a firm heading for restructuring. A cleaner approach weights two outcomes.

Shareholder control

Most valuations assume a non-controlling, financial investor and take management as it is. The emergence of a controlling investor, a strategic or activist holder, can signal change. When activists lay out a restructuring plan to win a shareholder vote, an analyst can recast the proposed operating, investing, and financing steps as a disaggregated forecast for the scenario in which the plan succeeds.

The pieces come together in a step-by-step valuation. The process runs: (1) collect the historical statements; (2) build the income statement projection; (3) build the balance sheet projection; (4) derive the free cash flow metrics from those projections; (5) set a terminal value from the model outputs; and (6) discount the flows and terminal value to today. Steps 2 and 3 are where the real work sits, since they generate the inputs for the cash flow, and the forecast narrative shapes the terminal value.

Folium AG is a hypothetical, publicly traded European firm in a mature, cyclical industry that has found an attractive growth niche. Its sales grew from EUR100 million to EUR140 million over five years, but net income slid from EUR12 million to EUR7 million.

An analyst believes the market is too pessimistic about Folium turnaround and that the stock, near a EUR200 million market capitalization, may be undervalued. The base-case assumptions, viewed as most likely, span a five-year horizon: the narrative is a profitability turnaround starting in year t+3, with disproportionate asset investment in t+1 to support it. Expenses grow faster than revenue in the first year (compressing margins) and slower thereafter (expanding them). All financing surpluses go to cash; no dividends are paid.

Because expense growth exceeds revenue growth only in t+1 and trails it in the last three years, margins dip first and then improve. Because asset growth tops revenue growth in t+1 and falls below it later, investment is front-loaded. These assumptions produce the forecast income statement and balance sheet below.

From these statements the analyst computes FCFF and FCFE. Recall that NOPAT is after-tax operating profit (revenue minus COGS, SG&A, and D&A, then taxed), and FCFF then deducts net investment.

The front-loaded investment holds free cash flow low in t+1 and t+2, then improving profitability lifts FCFF sharply, from a base year (t) FCFF of EUR6.5 million to EUR16.92 million by t+5. The analyst values Folium on FCFF, treating the zero-debt assumption as a weakness in the FCFE model, and estimates terminal value with a constant growth model (a suitable peer group for the niche is hard to define).

Finally, discount the FCFF stream and the terminal value, then subtract the market value of debt to reach equity value.

A single point estimate hides the range of futures a company might face. The analyst therefore re-runs Folium under alternative scenarios. An optimistic case assumes higher revenue growth, greater margin expansion, and heavier investment. A downside case has revenue growth stalling, margins continuing their recent slide, and investment cut to protect free cash flow. A fourth stable case, added after the base case was questioned as too rosy, grows expenses and assets at the same rate as revenue, so margins never improve. The scenarios spread the results widely.

Comparing the base and stable cases isolates the value of the turnaround: nearly all of the base-case upside comes from lifting EBIT margin above the 10.71 percent stable level. The downside case even pushes equity value slightly below zero, a signal of possible distress; in reality the market value of debt would fall so that equity stays just positive. To reach a single figure, the analyst assigns subjective probabilities, 60 percent to the base case, 20 percent to stable, and 10 percent each to the optimistic and downside cases, and takes the weighted average.

A second analyst argues the assumptions are too coarse and builds an expanded model that separates COGS, SG&A, and D&A; splits working capital from net PP&E; sets a 46 percent target debt ratio; spreads any financing surplus across cash (20 percent), debt (0 percent), and equity (80 percent); and specifies borrowing and cash interest rates. With a 10 percent cost of equity, this richer model values Folium equity at about EUR219 million, roughly 11 percent below the simpler version, a reminder that added structure can move the answer.

When only a disaggregated model will do

Folium is mature enough to have shown positive historical free cash flow, so a base-year growth approach was at least conceivable. A young, fast-growing firm with negative free cash flow leaves no such option. Consider Veranta Inc., a recently listed company whose revenue jumped from USD70 million to USD190 million in two years and might reach USD700 million within five, all while posting negative profits and cash flow. A base-year cash flow cannot be grown into a value when it is negative, so a disaggregated model that forecasts the path to positive cash flow is the only workable present value route. Its terminal value should come from a multiple of forecast year t+5 sales or earnings, not a constant growth model, because Veranta is still growing 20 percent a year with improving margins and has not reached the stability a constant growth model assumes. Applying a multiple to year t+1 earnings would fail twice over: those earnings are still negative, and a year t+1 multiple yields a current value, not a terminal one.

Across all of these, the strength of a disaggregated model is granularity: revenue drivers by line of business or region, specific cost drivers such as labor and raw materials, and efficiency drivers such as receivables or inventory management. That detail is what lets an analyst run sensitivity analysis, gauging the effect on equity value of moving one variable while holding the rest fixed.