Lesson 1 of 7 · All lessons in this module
- What to forecast: forecast objects and principles
- Forecast approaches and the forecast horizon
- Forecasting revenue: top-down, bottom-up and non-recurring items
- Forecasting operating costs: COGS, SG&A and segment margins
- Working capital, capital investments and capital structure
- Integrated case: fixed assets class by class and a marketplace cost base
- Scenario analysis: risk factors, cost structure and sensitivityVideo lesson
Free lesson: the first lesson of every module is open to everyone.
Lesson 1 of 7 · 12 min
What to forecast: forecast objects and principles
Before forecasting anything, choose the object: a driver, a single line, a summary measure or an ad hoc item, preferring objects the company discloses regularly and keeping the model only as detailed as accuracy requires.
In short
- Four forecast objects: drivers of financial statement lines, individual lines, summary measures (FCF, EPS, total assets) and ad hoc objects not yet in the statements.
- Forecasting drivers (e.g. stores × sales per store) gives the most explanatory value and can improve accuracy, because drivers inside one line may move in opposite directions.
- Summary measures are efficient but less transparent, so the forecast is hard to audit; use them when the measure is stable or disclosures are thin.
- Prefer objects that are disclosed regularly (or can be computed from what is), so forecasts can be verified against actual results in good time.
- Avoid overly complex models: more objects mean more work to build and update, often with no gain in accuracy.
Four kinds of forecast objects
A forecast object is the thing the analyst actually projects. The choice depends on the information available and on the trade-off between efficiency, accuracy, explanatory value and verifiability.
Drivers of financial statement lines are the quantities that produce a line. A warehouse retailer's net sales can be built from the average number of stores times average sales per store; cost of sales and SG&A can then be forecast as percentages of that net sales figure. Drivers explain *why* a line changes, and they help when the drivers move in different directions, which is hard to forecast on an aggregated basis.
Individual financial statement lines are forecast directly when there is no clear driver, when the item is small, or when the analyst has no particular view on it. Examples are amortization expense (management often gives an estimate), a small 'other non-current assets' line, or thinly disclosed cash flow lines. A common shortcut is to hold the item at its past level.
Summary measures such as free cash flow, EPS or total assets are forecast as a whole. Ad hoc objects are items not yet in the historical statements, such as the cost and timing of a newly announced lawsuit, a regulatory action, a tax dispute or a natural disaster, which the analyst may need to estimate before the company books any accrual.
| Object | Example | Strength | Weakness / when to use |
|---|---|---|---|
| Drivers | Stores × sales per store; gross margin % | Explanatory value, often better accuracy | Needs disclosure of the drivers |
| Individual lines | Amortization expense, small other assets | Quick for minor or driverless items | Little insight; used for immaterial lines |
| Summary measures | FCF, EPS, total assets | Efficiency: few variables | Less transparent, hard to audit; best if stable or disclosure is limited |
| Ad hoc objects | Pending lawsuit, tax dispute, disaster loss | Captures events not yet accrued | Requires judgment on size and timing |
Why drivers beat the aggregate line
A line item can hide drivers that pull in opposite directions. If a retailer opens 5% more stores while sales per store fall 3%, net sales still grow, but only about 1.85%, because the two effects compound: 1.05 × 0.97 − 1 = 1.85%.
An analyst who forecasts net sales directly at, say, last year's 5% growth would miss the falling productivity of the stores. Forecasting the two drivers separately shows the problem and lets the analyst ask the right question: why are sales per store falling?
Principles: disclosed regularly, no more complex than needed
Use objects that are disclosed regularly or can be calculated directly from regular disclosures: net sales, store counts, sales per store, or a government's monthly national retail sales series. Data that appears only occasionally, such as a consultancy's market-size report, is useful *background* but a poor forecast object, because the forecast cannot be confirmed in a timely way.
Verifiability matters at every level of detail. If a company reports gross margin only for the whole group, forecasting gross margin by product line cannot be checked: when the reported margin misses the forecast, which product was wrong? Likewise, a quantity disclosed only annually can be checked only once a year.
Be wary of overly complex models. A drug company that reports sales for dozens of products by region could in theory be modelled product by product and illness by illness, but the model would take months to build, weeks to update each quarter, and many of its inputs could never be verified. Focus on the drivers that matter most and use aggregates and shortcuts where detailed forecasting adds little value, such as small or mature products.
A useful test: if the company reports this number every quarter, a forecast of it gets graded every quarter. If it never reports it, the forecast never gets graded, and errors can hide for years.
Formulas
- Driver-based revenue (capacity example)
Growth in the line = (1 + growth in stores)(1 + growth in sales per store) − 1.
- Costs from margin drivers
Worked examples
Example 1 Building net sales and costs from drivers
Brightwell Clubs plc expects to operate an average of 220 warehouse stores next year, with average net sales per store of GBP 48 million. Gross margin is forecast at 12.5% and SG&A at 9.0% of net sales. What are forecast net sales, cost of sales and SG&A?
Example 2 Naming the forecast object
An analyst covering a regional bank (1) forecasts total assets directly at last year's growth, (2) takes next year's amortization expense from the annual report note, (3) estimates a fine from a newly opened regulatory investigation, and (4) forecasts loans × average loan yield. Classify each object.
Example 3 Should the analyst forecast by region?
A packaging company reports revenue for four geographic regions, but all four serve the same customers and face the same economic conditions. Gross margin is disclosed only for the group. Should the analyst build separate regional revenue and margin forecasts?
On the exam
Expect concept items: classifying a forecast object, the benefit of summary measures (efficiency) versus drivers (explanatory value), and why objects should be regularly disclosed.
- Saying summary measures are more transparent; they are more efficient but harder to audit.
- Treating a pending lawsuit or tax dispute as an ordinary line item; it is an ad hoc object not yet in the statements.
- Using one-off third-party reports as forecast objects; they inform forecasts but cannot be verified in a timely way.
- Assuming more detail always means more accuracy.
Check yourself
Q1A retailer is expected to average 150 stores next year with sales per store of EUR 40 million and a gross margin of 15%. Forecast cost of sales is closest to:
Q2A benefit of using summary measures such as EPS as forecast objects is most likely:
Q3Which of the following is best described as an ad hoc forecast object?
Q4A company discloses gross margin only on a consolidated basis. An analyst considering a product-line gross margin forecast should most likely:
Q5A retailer's store count is forecast to rise 4% and sales per store to fall 2%. Forecast growth in net sales is closest to:
Answer the 5 questions: +10 XP for each one you get right first time.
Liked this lesson? Get the rest free for 7 days
Create a free account to open the other 6 lessons in this module, with videos, flashcards that come back before you forget them, the question bank and full mocks. No card needed.