Internal transfer pricing

Why do two divisions of the same company argue bitterly over a price when the money never leaves the company?

Two divisions of one company are locked in a months-long dispute over what one should charge the other for a component. Take the shareholder's view. Every pound at issue leaves one internal account and lands in another; the consolidated accounts are identical whatever they settle on. The argument looks like pure waste, and the people conducting it like they have forgotten who they work for.

They have not. Something real is at stake, and finding out what tells you more about how large firms work than the accounting rule ever will.

Begin with why the price exists at all. A firm exists partly because directing beats contracting — inside its boundary the price mechanism is deliberately switched off and people are simply instructed. But switch prices off across a company of forty thousand people and you lose the one thing prices did well: they told you, cheaply, whether an activity was worth doing. So large firms carve themselves into divisions with their own profit statements and reinstall an internal price to keep that signal. The dispute is the cost of that reinstallation.

Now ask what the internal price is being asked to do. Three jobs, and they pull apart. Decision: it should tell the buying division whether to buy inside, buy outside, or not at all, in a way that matches what is best for the whole firm. Motivation: divisional managers act on their own reported profit, so the price must leave each side wanting the right thing. Measurement: head office wants divisional profit to be a fair reading of divisional performance, for promotion and bonus.

Watch one number fail all three. The upstream division makes a part at a variable cost of 60 and carries fixed costs allocating at 20, so full cost is 80. The downstream division adds 30 of its own cost and sells for 100. For the firm, each unit brings in 100 and consumes 60 + 30 = 90 of real cash cost, so it earns 10. Make it.

Set the transfer price at full cost, 80 — the intuitive "fair" answer. The downstream manager now sees 80 + 30 = 110 against a selling price of 100 and declines the business, costing the firm 10 a unit. The allocated 20 is not avoided by refusing; it is incurred either way. The manager was not obtuse — they optimised the number they are paid on.

Set it at variable cost, 60. The decision is now right — 60 + 30 = 90 against 100, so the division buys and the firm earns its 10 — but the upstream division books zero margin and appears at review to be a cost centre that never earns anything. Its manager will chase external orders and will not invest in a customer that pays them nothing.

So which price is correct? The one that resolves the decision job is variable cost plus the opportunity cost of the capacity displaced. With spare capacity that is zero, so the right price is 60. If the upstream division is full and could have sold the unit outside at 100, the displaced contribution is 100 - 60 = 40 and the right price is 60 + 40 = 100, the market price. The correct transfer price therefore moves with capacity utilisation, week by week — and no manager accepts a price that swings against them for reasons outside their control, so a rule nobody accepts is not a rule.

Which is why real firms use blunt instruments instead — market price where a genuine external market exists, full cost, cost plus a markup, or negotiation. Negotiation is the revealing one: it hands the divisions the bargaining problem the firm was created to avoid, and rewards the better negotiator over the better operator. Head office can impose a price and end the argument, but divisional profit then stops being a performance measure at all, because a chunk of it was assigned by decree.

If one number really is doing incompatible jobs, there is a test. Firms that stop paying divisional managers on divisional profit — evaluating them on cost, quality and delivery instead — should see internal price disputes shrink to a clerical matter. If the arguments were just as bitter there, the driver would be something else, such as status or headcount, and this account would be wrong.

First, correct the framing that a zero-sum transfer is irrational to argue about. It is zero-sum in the consolidated accounts and not at all zero-sum in the managers' payoffs — and it is the managers who make the operating decisions. Any account treating the firm as a single decision-maker gets this wrong every time.

There is a fix that almost works, called dual pricing: charge the buyer variable cost so its decision is right, credit the seller market price so its motivation is right, and let head office absorb the difference. It repairs the first two jobs and breaks the third, since divisional profits now sum to more than the firm earned. That the honest fix is the one firms rarely adopt tells you which of the three jobs they care most about.

One more force distorts everything. Where the divisions sit in different countries the transfer price also decides which tax authority sees the profit, which is why tax law imposes an arm's-length standard on it. When the tax answer and the management answer disagree the tax answer usually wins, and the internal price becomes a compliance number managers are then asked to run a business on.

• Why some firms abandon divisional profit entirely and run internal units as cost centres with service levels.

Every term the collection defines is gathered in the glossary.