FSA 12 Introduction to Financial Statement Modeling

Experts across many fields make persistent forecasting errors that trace back to behavioral biases, and analysts building financial statement models are not exempt. Awareness of the biases and their remedies improves both forecasts and the decisions built on them. Five biases matter most here: overconfidence, conservatism, the illusion of control, representativeness, and confirmation bias.

Overconfidence

Overconfidence is unwarranted faith in one’s own ability. In one classic study, forecasts stated with 90 percent confidence, which should be wrong only 10 percent of the time, turned out wrong as often as 40 percent of the time. Overconfidence also tends to rise when people make contrarian calls that run against the consensus. The remedy is to record and share forecasts and review them regularly, noting both hits and misses; most analysts find they are wrong at least as often as they are right. Since error rates are high, the useful response is to widen the confidence interval, and scenario analysis is one direct way to do that. Asking where the forecast could be wrong, and by how much, generates a range of outcomes rather than a single point.

Illusion of control

Closely linked to overconfidence, illusion of control is the tendency to overestimate how much one can control the uncontrollable and to take fruitless steps in pursuit of control. In modeling it shows up as a belief that accuracy improves by gathering ever more information and expert opinion and by building ever more granular models. Both help only up to a point. Material information is finite, so adding immaterial detail misleads; complex models tend to be overfitted to history and prove fragile against outliers; excessive breadth hides assumptions and makes updating harder; and the extra hours carry an opportunity cost of fewer companies examined. The remedy is to restrict variables to those the company regularly discloses, focus on the highest-impact drivers, and speak only with those likely to offer a genuinely distinct perspective.

Conservatism

Conservatism is the failure to update a prior view enough when new information arrives. Most often it appears as reluctance to absorb bad news such as a weak earnings print or a competitor’s move, though it can also mean underweighting good news and leaving estimates too low. Another name for it here is anchoring and adjustment: the prior estimate acts as an anchor, and the adjustment away from it is too small, so the revised figure stays too close to the old one. Remedies include regular team reviews of the model, for instance each quarter, and keeping models flexible with fewer variables so assumptions are easy to change.

Representativeness

Representativeness is the habit of classifying new information by resemblance to familiar categories, even when the new case is genuinely different. In forecasting, its common form is base-rate neglect: ignoring how a phenomenon behaves across the larger population (the base rate, or outside view) in favor of case-specific detail (the inside view). Modeling a biopharmaceutical company purely from its own drugs and salesforce is the inside view; using industry averages for gross margin and research spending is the outside view. Neither alone is superior; the strongest forecast uses both, typically starting from the base rate and then adjusting for the specific factors that make the target company differ from its class.

Confirmation bias

This bias is the habit of seeking out and noticing whatever supports a prior belief while discounting whatever contradicts it. Among analysts it shows up as a research process aimed only at positive news, or built with too narrow a scope: studying a company thoroughly but its competitors and substitutes only lightly, or speaking mainly to management and other bulls who confirm the existing view. One striking illustration is a word count run on a major European bank’s 2007 annual report, released just months before the government rescued it: positive words hugely outnumbered negative ones. Management should still be part of the research process, but their built-in optimism must be recognized. The remedies are to read or speak with analysts holding a negative view and to seek out colleagues who are not economically or emotionally invested in the security.

Competition shapes both the prices a company can charge and the input prices it must pay. Michael Porter’s five forces framework, covering the threat of substitutes, rivalry, supplier power, buyer power, and the threat of new entrants, gives analysts a structured lens on how competition shapes financial outcomes. A closely related profitability measure is return on invested capital (ROIC), an after-tax figure equal to net operating profit after adjusted taxes, divided by invested capital, which itself is operating assets net of operating liabilities. High and persistent ROIC often signals a durable competitive advantage.

A favorable structure: the cognac industry

Cognac, Remy’s most important segment at over 90 percent of total operating profit, sits in a structurally attractive market: supply is constrained and demand is rising. Production is tightly regulated by The Bureau National Interprofessionnel du Cognac, allowed only within a restricted zone surrounding the town of Cognac, in southwestern France, with annual volume capped. About 98 percent of output is exported. The market is very concentrated: the four leading houses hold 78 percent of world volume and, by value, 84 percent of the global total, while Remy holds roughly 16 percent of volume and 18 percent of value. Demand has grown on Asian strength, particularly China and Singapore, more than offsetting a softer European market, and the global spirits market grew more than 5 percent a year over 2000 to 2017. Consumers have also traded up toward pricier, higher-quality cognac, improving mix.

Every force points toward strong and defensible profitability, which supports the model’s assumption of high and improving Cognac margins. Porter draws a distinction between the five forces and other influences such as government and taxes: government is not a sixth force, because state action is not by itself either helpful or harmful to profitability. The right way to read government influence is to trace how a specific policy works through the five competitive forces.

The same company, different structures

Competitive structure can differ sharply by country for one company, changing the revenue, margin, and capital forecasts an analyst would make. Anheuser-Busch (AB) InBev, the biggest brewer in the world, meets very different conditions in Brazil and the United Kingdom. In Brazil, its subsidiary AmBev dominates with about 65 percent share, against Heineken near 20 percent and Petropolis near 12 percent, and reported an EBITDA margin near 50.4 percent in 2018, the highest in the global beer industry; competition centers on distribution and premiumization rather than price, so an analyst would forecast solid revenue growth and continued above-average profitability. In the United Kingdom, the same four-player market is fragmented, with Heineken around 24 percent and MolsonCoors, AB InBev, and Carlsberg near 18, 18, and 11 percent, so there is no dominant brewer. Declining volumes, high fixed costs, a consolidated and powerful retail customer base, and a shift from on-trade to off-trade drinking make price competition intense and operating margins below 10 percent, so an analyst would forecast only cautious revenue growth, if any.

Competition, pricing, and cost forecasts

The next example works through how a tax shock, elasticity, and cost structure combine to move margins, using a fictional brewer, EuroAlco, the market leader in a growth market with 35 percent share. Its government tripled the beer excise duty from EUR0.3 to EUR0.9 per liter and announced a further EUR0.1 increase. Both analysts covering the company assume the further excise increase is passed to consumers as a 10 percent retail price rise, that the brewer’s net price per hectoliter is unchanged, that half of cost of sales is fixed and half varies with volume, that selling expenses hold at a constant share of sales, and that administrative expenses do not move.

Return on invested capital ties these competitive pressures to value. In a later EuroAlco scenario, larger retailers strengthen buyer power. One analyst sees operating margin falling from 13 percent to 8 percent with invested capital unchanged at EUR3,000 million, so ROIC (pre-tax operating profit over invested capital) drops from about 27 percent to 16.6 percent. A second analyst assumes the company holds margin at 13 percent by offering longer credit terms, but invested capital doubles to EUR6,000 million, so ROIC falls to 13.5 percent. The lower-margin, capital-light path is the better outcome for the company. Porter’s framework helps an analyst judge whether sales growth and margins are likely to run high or low relative to history, the economy, and rivals, but the link from structure to a specific forecast is a matter of judgment, not a mechanical rule.

General increases and decreases in prices can materially distort forecasts of revenue, profit, and cash flow. The impact differs across companies, and within a single company it usually differs between revenue and expenses. The pivotal question is pricing power: how much of a rise in input cost a company can pass on through higher selling prices. A powerful brand (Coca-Cola) or proprietary technology (Apple) supports pass-through, and companies that can pass costs on tend to earn higher, steadier profit and cash flow.

Sales under inflation and deflation

Rising input costs eventually feed into higher end-product prices, but industry structure governs how completely and quickly. In the United States, beer is an oligopoly in which AB InBev holds nearly half, and a three-tier system forces brewers to sell through wholesalers rather than straight to large retailers, fragmenting the customer base. That structure has let US brewers raise prices roughly in line with the Consumer Price Index. European brewers, selling through concentrated retail chains such as Carrefour, Tesco, and Ahold and lacking a dominant national player, have weaker pricing power, and their price changes have averaged about 100 basis points below customer inflation.

Passing inflation through raises a volume risk when demand is price elastic, which it is when cheap substitutes exist. If a company could lift prices 10 percent while holding unit volume to offset a 10 percent input cost rise, gross margin percentage would be unchanged and absolute gross profit would rise. In the short run, though, volume usually falls after a price increase, by an amount that depends on elasticity, competitor reactions, and substitute availability. The trade-off runs both ways: in inflation, raising prices too late squeezes margin while raising too early loses volume; in deflation, cutting too soon lowers margin while waiting too long loses volume. The first mover on a price cut typically wins a short-lived volume bump before rivals match.

An analyst can use such a relationship directly: assume a level of inflation, read off the implied volume growth, and compute revenue. For an international company, the pricing assumption should reflect the geographic mix, since inflation rates differ by country, and strategy and competition also shape prices. AB InBev illustrates the currency angle: its Latin America South unit posted organic revenue growth of 24.7 percent in 2011, with volume adding only 2.1 percent and price the remainder, and organic operating profit up 27 percent. High inflation in an export market normally pressures that country’s currency, so pricing gains can be wiped out by currency losses; the absence of a currency hit for AB InBev that year was a positive surprise, not the norm. Most analysts handle high foreign inflation by assuming a normalized constant-currency growth rate after a year or two, which can understate near-term revenue; others try to model the currency effect directly, which is imperfect given the difficulty of projecting exchange rates.

Strategy is decisive. Facing rising input costs, a company can protect margin by passing costs on, or it can absorb some margin loss to win market share by not fully raising prices. Sysco, the largest North American food distributor, has at times declined to pass on food-price increases in recessions to avoid weakening already-stressed customers such as restaurants and schools. In contrast, the big French cognac houses raised prices sharply in China in 2011 and 2012 specifically to cool strong demand, since aging cognac for later sale at higher prices can beat maximizing near-term volume.

Costs under inflation and deflation

Forecasting costs benefits from knowing an industry’s purchasing habits. Long-term fixed-price contracts and hedges delay the effect of input price moves, so an analyst covering an industry that uses them would let cost changes flow through more slowly. Monitoring the underlying drivers helps too: weather swings the price of agricultural inputs, so spotting a weather pattern can feed a cost forecast. The competitive environment also matters. A company with alternative inputs or vertical integration can dampen input volatility. The coffee firm JDE, for instance, blends different beans and can shift the mix to hold taste and cost steady, but if every supplier country raises prices at once, blending fails and it must either raise prices or, if rivals such as Nestle make that hard, cut advertising, which helps short-term profit but can weaken the brand over time. Shared inputs can also spread a shock: after Russia’s 2010 heat wave destroyed grain, Carlsberg paid more for Russian barley and, by importing from Western Europe, pushed up barley prices there too.

For company costs, it helps to segment the cost base by category and geography and to assess each item’s exposure to inflation, including the scope to substitute cheaper inputs or raise efficiency. In 2011, a jump in raw material prices cut the gross margins of Unilever and Nestle by 110 to 170 basis points, yet efficiency gains such as trimming advertising let both lift overall operating margin that year. The next example uses common-size (percent-of-sales) analysis to show why the same cost inflation hits two companies differently.

How far out to forecast is a deliberate choice, shaped by four factors: the investment strategy the security is being weighed for, the cyclicality of the industry, company-specific events, and the analyst’s employer. Most managed strategies state an average holding period, which should match average annual portfolio turnover, computed as one divided by turnover. A three-to-five-year horizon, for example, implies annual turnover between 20 and 33 percent. Cyclicality argues for a horizon long enough for the business to reach a mid-cycle level of sales and profit, and company events such as a recent acquisition or restructuring argue for enough time to let the expected benefits show up in the statements. Sometimes the employer simply fixes the parameters.

Longer horizons often capture normalized earnings better than short ones. Normalized earnings are the expected mid-cycle earnings absent any unusual or temporary factor, positive or negative, such as the business-cycle stage, recent mergers, or restructuring. Normalized free cash flow is defined similarly, as mid-cycle operating cash flow adjusted for those unusual items less recurring capital expenditure. Extending the forecast lets the analyst strip out temporary effects and estimate what the company would earn in a normal year. As with any projection, a long-term forecast starts from revenue, and the growth-relative-to-GDP and market-growth-and-share methods used for short horizons apply equally to long ones.

Estimating normalized revenue

A common technique fits a trend line to historical revenue and reads normalized figures off that line. For Continental AG, a global automotive supplier, a least-squares regression of revenue on time (the TREND function in a spreadsheet) produces a rising straight line; actual revenue oscillates above and below it, dipping far below in 2020. The trend value for each year, extended forward, gives a normalized revenue path that smooths out the year-to-year swings.

Terminal value and inflection points

A DCF model needs a terminal value to capture going-concern value beyond the explicit forecast, and several judgments shape it. First, weigh whether the terminal-year free cash flow needs normalizing: if it is unusually low for cyclical or capital-spending reasons, an adjustment may be warranted. Second, consider whether the long-term growth rate should differ from history, since even a mature company can reaccelerate through innovation or expansion (Apple) while a seemingly safe grower can decline on technological change (Eastman Kodak). Because most DCF models use a perpetuity that grows the final-year cash flow at a constant rate forever, and because the perpetuity can be a large share of total value, the cash flow used must be a normalized, mid-cycle figure; starting a perpetuity from a boom year badly overstates value, and a trough year understates it.

The hardest task is anticipating inflection points, when the future breaks sharply from the recent past. Sudden economic disruption, such as the 2008 crisis or the pandemic, can throw even a sound company off course, especially a highly leveraged one. Regulation and technology are the other main drivers: government actions can hit some businesses hard and fast, and technology can render a fast-growing innovator obsolete within months. Exposure varies by industry, utilities face heavy regulation but slow technological change, semiconductor makers the reverse, and pharmaceutical companies both. Finally, the long-term growth rate is a key perpetuity input: some firms (Tencent, Amazon, Google) can outgrow the economy for long stretches, while others (print media) grow slower or shrink, and an unrealistic growth assumption pushes the whole valuation off the mark.