A pottery sells mugs for 12 each. The accountant's sheet says each mug costs 9 to make, so when a café offers 7 apiece for a one-off batch, the owner refuses: it would lose money on every mug. The accountant's sheet is right about the average and wrong about the decision. Five of those 9 are clay, glaze and gas, which are spent only if a mug is made. The other 4 are rent and the kiln's depreciation, which are spent whether the café's mugs are made or not. If the kiln sits idle on Tuesdays, the batch adds 2 per mug to the day's profit.
That gap, between the cost on the sheet and the cost that changes with the choice, is most of managerial accounting. A graduate of the course carries one reflex for life: before using a cost, ask what decision it is for. The same factory has a different "cost" for pricing, for a make-or-buy choice, for judging a manager and for the tax return.
Not advice. This explains how internal costing works. Companies and numbers in the examples are invented. It is not accounting, tax or investment advice.
- Who takes it
- Accounting, finance and business majors, usually in year 2 after the first financial accounting course, and many engineering-management and operations students. Business-school cores, the professional bodies (CIMA in the UK, the IMA's CMA in the US, ICAEW and ACCA) and the standard textbooks all cover the same ground: cost behaviour, product costing, budgets and variances, decision analysis and performance measurement.
- What it is for
- Giving managers numbers they can act on, rather than numbers an outsider can audit.
- Comes after
- Accounting (34/50), including its break-even section, which this piece takes further.
- Leads to
- Operations Management (34/52), Strategy & Competition (34/54) and Entrepreneurship (34/56); inside this major, Taxation (36/48) borrows its transfer-pricing idea.
- What this piece gives you
- You will not be able to build a costing system for a plant. You will see why fixed costs are a lever, why profit can be raised by building stock, why a "loss-making" product may be nothing of the sort, and how to read a variance. Easier first reads: Accounting (34/50) and Pricing Without Fear (04/03).
Cost behaviour: fixed costs are a lever
Costs behave in only a few ways. A variable cost rises with activity: clay, card packaging, a sales commission. A fixed cost does not move over the range you care about: rent, a salaried supervisor, a lease. Plenty are in between (a step cost jumps when you hire the next person). The split matters because of one number, the contribution margin: price minus variable cost per unit. It is what each extra unit contributes toward covering fixed costs, and then toward profit.
The first course in accounting stops at break-even. The managerial course asks the next question: how sensitive is profit to sales? That depends on the mix of fixed and variable costs. A firm with large fixed costs has a high operating leverage: a small change in sales becomes a large change in profit, in both directions. The degree of operating leverage is contribution ÷ profit, which equals units ÷ (units − break-even units).
Both firms sell at 40. Hand-crafted: variable cost 32, fixed 24,000. Automated: variable cost 16, fixed 60,000. Invented numbers.
Degree of operating leverage is undefined at break-even (profit zero) and shown as "n/a" there or below it.
Without JavaScript: at 4,000 units, hand-crafted earns 8,000 (leverage 4.0); automated earns 36,000 (leverage 2.7). They cross at 2,250 units. Below that the hand-crafted firm loses less; above it the automated firm wins by more.
The automated firm breaks even sooner (2,500 units against 3,000) and earns much more when sales are strong, but it loses 12,000 at 2,000 units where the hand-crafted firm loses 8,000. Neither is "better". A graduate sees the cost structure as a bet on volume, which is why airlines, hotels and software firms (mostly fixed costs) are so jumpy in a downturn.
Absorption versus variable costing: profit you can make by producing
To value stock on the balance sheet, accountants must decide whether a unit made in the factory carries a share of the factory's fixed overhead. Absorption costing says yes: every unit gets direct materials, direct labour, variable overhead and a slice of fixed overhead. Variable (direct) costing says no: fixed overhead is a cost of the period and is expensed in full. External reporting rules require absorption: IAS 2 under IFRS, and ASC 330 under US GAAP, both allocate fixed production overhead to inventory (IFRS on the basis of normal capacity). Variable costing is a tool for managers.
The two give the same profit when production equals sales. They diverge when stock changes. Under absorption, a unit left in the warehouse carries its slice of fixed overhead with it onto the balance sheet; it does not hit profit until sold. So a manager can raise reported profit simply by making more than is sold.
Sells at 50; variable production cost 20; fixed factory overhead 60,000; normal capacity 10,000 units. The firm sells 8,000. Invented numbers.
Without JavaScript: producing 8,000 gives absorption profit 180,000. Producing 10,000 gives 192,000, and 14,000 gives about 205,714, though the same 8,000 units are sold and cash is poorer. Variable-costing profit stays at 180,000.
This is a standing incentive problem: if a bonus is tied to absorption profit, stock-building pays. Lean manufacturing's push toward low inventory led many firms to report variable-costing profit internally for exactly this reason. The profit "earned" by stock is real in the accounts and unreal in cash: the extra units cost money to make and are not yet sold.
Overhead and activity-based costing: who really causes the cost
Direct costs trace to a product (the clay in a mug). Overhead does not: supervision, set-ups, inspection, the cost of changing the line from one product to another. It has to be allocated, and the traditional way is a single rate based on volume, such as direct labour hours. That is cheap to run, and misleading whenever products differ in how much support they need.
Activity-based costing (ABC) was popularised by Robin Cooper and Robert Kaplan in the late 1980s, building on H. Thomas Johnson and Kaplan's critique Relevance Lost (1987). It splits overhead into pools by activity (set-ups, machine time, inspections), finds a driver for each, and charges products by how many activities they actually use. The classic finding: low-volume, special-order products are under-costed by the volume method and high-volume standard ones over-costed, so a firm happily takes custom work it loses money on.
Overhead pools (invented): set-ups 90,000, machine time 60,000, inspection 30,000. Standard: 10,000 units, 20 set-ups, 9,000 machine hours, 50 inspections, 10,000 labour hours. Custom: 500 units, 2,000 labour hours, 1,000 machine hours, 150 inspections. Direct cost 120 per unit for each; the custom unit sells for 300.
Without JavaScript: by labour hours, each custom unit gets 60 of overhead (total cost 180, profit 120). Activity-based with 70 set-ups, it gets 197 (total cost 317, a loss of 17 per unit).
ABC is not free. Surveying the activities, measuring the drivers and keeping it current is heavy work, which is why Kaplan and Steven Anderson proposed time-driven ABC in 2004, using the minutes each activity takes. Critics also note that it still allocates costs that do not change with the decision, so it informs pricing and mix but does not replace relevant-cost analysis, the next idea.
Relevant costs and the bottleneck: only what differs counts
A cost is relevant to a decision only if it will differ between the options and lies in the future. Two consequences are famous. Sunk costs (the machine already bought, the research already done) are not relevant, because no option can recover them. And allocated fixed costs that continue regardless are not relevant either. Which brings us back to the mug.
A café wants 2,000 mugs. Normal price 12; variable cost 5; fixed cost 4 per mug at normal volume. Spare kiln capacity is unknown. Each mug that displaces a normal sale loses its contribution of 7.
Without JavaScript: at 7 a mug with 2,000 spare, the order adds 4,000 to profit even though full cost is 9. With no spare capacity, it costs 10,000, because 2,000 normal sales worth 7 each are lost.
The second famous idea is the bottleneck. When one resource limits output, the product to push is not the one with the biggest margin per unit, but the biggest margin per unit of the scarce resource. Eliyahu Goldratt dramatised this in the novel The Goal (1984), and it is the heart of his Theory of Constraints. (With several constraints at once the maths becomes a linear program.)
One machine limits the plant. Three products (invented): A margin 20, 4 machine minutes, demand 1,000. B margin 12, 1.5 minutes, demand 1,500. C margin 40, 10 minutes, demand 400.
Without JavaScript: C has the biggest margin per unit (40) but the smallest per minute (4). B earns 8 a minute, A 5. With 6,000 minutes the best plan makes 1,500 B, 937 A and about 0 C, for 36,740.
Budgets and variances: compare to the flexed budget
A budget is a plan in numbers. A variance is the gap between plan and actual, and the managerial skill is in reading what caused it. The beginner's mistake is to compare actual cost to the static budget, built for a different volume. If the firm made more than planned, it is bound to spend more. The fix is the flexed budget: what costs would have been at the actual volume, given the standards.
Then the gap splits into two questions you can put to two different people. The price variance is (actual price − standard price) × actual quantity: did we pay more for the material? The efficiency (usage) variance is (actual quantity − standard quantity allowed) × standard price: did we use more than we should have? Purchasing answers the first and the factory floor the second.
Plan: 900 mugs, 3 kg of clay each, standard price 4.00 per kg. Actual output: 1,000 mugs. Set what the clay actually cost and how much was used.
Without JavaScript: at 4.20 and 3,100 kg, actual cost is 13,020 against a static budget of 10,800 (2,220 over). Against the flexed budget of 12,000 it is 1,020 over: 620 from price and 400 from usage.
Budgets also shape behaviour. If next year's budget is this year's spending plus a bit, managers pad estimates and spend the remainder in the last quarter. Critics (the "Beyond Budgeting" movement of Jeremy Hope and Robin Fraser) argue for rolling forecasts and relative targets; defenders say budgets still do the work of coordination and authority that nothing else does. Firms differ, and the field has not settled it.
Transfer prices and the measures that steer people
When one division of a company sells to another, the transfer price is revenue for the seller and cost for the buyer, and it nets to zero for the company. It matters anyway, because divisions are judged on their own profit. Good transfer pricing makes the division's best choice match the company's best choice. The textbook rule: the seller's minimum is its variable cost plus whatever it gives up by selling inside (the lost outside sale if it is at capacity); the buyer's maximum is the lower of what the item is worth to it and what it could buy outside.
Division S makes a part at variable cost 30 and can sell it outside at 48. Division B can buy outside at 45 and values the part at 52 in its own product. Invented numbers.
Without JavaScript: with spare capacity, S's minimum is 30 and B's maximum is 45, so any price from 30 to 45 makes the trade happen, adding 15 per part to the company. At full capacity S's minimum is 48, above B's 45, so no trade is right: it would cost the company 3 per part.
Across borders, the same idea becomes a legal question, because the price decides which country sees the profit. Tax authorities apply the arm's-length principle from the OECD Transfer Pricing Guidelines: intra-group prices should resemble what independent parties would agree. Taxation (36/48) returns to this.
The wider lesson is that every measure steers. Goodhart's law, in Marilyn Strathern's wording, says that when a measure becomes a target it ceases to be a good measure. Absorption profit rewards stock; cost-per-unit rewards long runs; division profit may reward hoarding capacity. Kaplan and David Norton's balanced scorecard (Harvard Business Review, 1992) answered by pairing financial measures with customer, internal-process and learning perspectives. It has been widely used, and also criticised for being a list without a causal model.
Outside the Anglo-American tradition the same questions have local answers. Germany's flexible marginal costing (Grenzplankostenrechnung) is a detailed variable-cost system; Japanese manufacturers popularised target costing, where the allowed cost is the market price less the desired margin, and the design team must hit it.
What a graduate just sees
- "What does it cost?" has no answer until you say what decision the number is for.
- Fixed costs are a lever on profit: high operating leverage means big gains in a boom and big losses in a slump.
- Absorption profit can rise when stock rises; check cash and stock before applauding profit.
- An allocated overhead is a convention. Low-volume, high-complexity products are usually under-costed by volume rates.
- Sunk costs and costs that continue anyway are irrelevant; opportunity costs, which never appear in the ledger, are not.
- With a bottleneck, rank products by margin per scarce minute, not margin per unit.
- Compare actuals to the flexed budget, then split the gap into price and usage.
- Any target changes the behaviour it measures.
Where it shows up next
- Operations Management (34/52): the bottleneck rule is the opening idea of throughput thinking and queues.
- Strategy & Competition (34/54): cost leadership depends on knowing true product costs and the experience curve; ABC is how firms find which customers lose money.
- Taxation (36/48): transfer prices decide which country taxes the profit.
- Entrepreneurship (34/56): unit economics, contribution margin and break-even are how a new product is judged before it has history.
- In the world: airline fare classes (selling spare seats at a price above the cost of one more passenger), hospital costing, a supermarket's decision to drop a product line, cloud-cost dashboards.
If you want to go deeper
- 📘 Textbook. Charles Horngren, Srikant Datar and Madhav Rajan, Horngren's Cost Accounting: A Managerial Emphasis (Pearson; the 17th edition appeared in 2021 and a newer one is listed), the usual choice for cost accounting. Ray Garrison, Eric Noreen and Peter Brewer, Managerial Accounting (McGraw Hill), is another widely used text, a little lighter on theory. Check for the current edition.
- 📘 Gentler. Eliyahu Goldratt and Jeff Cox, The Goal (1984), a novel about a failing plant that teaches the bottleneck idea better than any chapter.
- 🌐 Free text. OpenStax, Principles of Accounting, Volume 2: Managerial Accounting, free online at openstax.org.
Inside a firm there is no single cost. There is the cost for a decision: the part of the spending that will differ with your choice. Fixed costs make profit sensitive to volume, allocation rules decide which products look good, and the measures you choose will steer the people you measure.
Try this: pick something you pay for by subscription (a gym, a streaming service). Work out what one more use costs the provider (almost nothing) and what it costs you on average. Which number would you use to decide whether to cancel, and which would the provider use to decide whether to let you stay?
Sources. IASB, IAS 2 Inventories (allocation of fixed production overheads on the basis of normal capacity); FASB ASC 330 Inventory; Johnson and Kaplan, Relevance Lost: The Rise and Fall of Management Accounting (Harvard Business School Press, 1987); Cooper and Kaplan, "Measure costs right," Harvard Business Review (1988); Kaplan and Anderson, "Time-driven activity-based costing," Harvard Business Review (2004); Kaplan and Norton, "The balanced scorecard," Harvard Business Review (1992); Goldratt and Cox, The Goal (1984); Hope and Fraser, Beyond Budgeting (2003); Strathern, "'Improving ratings': audit in the British university system," European Review (1997) for the wording of Goodhart's law; OECD Transfer Pricing Guidelines for Multinational Enterprises and Tax Administrations. Every figure in the interactives was computed in node. Companies, products and amounts are invented. This describes costing methods and is not accounting, tax or investment advice.
- 📘 Horngren, Datar and Rajan, Cost Accounting: A Managerial Emphasis
- 📘 Goldratt and Cox, The Goal
- 🌐 OpenStax, Principles of Accounting, Volume 2: Managerial Accounting