Essay 02

Numbers, not voices

Firms were invented to end haggling. Once a year, over the money that matters most, most of them haggle anyway.

Every autumn, in thousands of companies, the same contest is staged. Division heads present their plans. Each argues that his unit is the future and the others are the past. The most persuasive, the most senior or the most feared walks away with more than last year; the rest settle for about the same. Nobody calls it a negotiation. It is one.

That is odd, because by one celebrated account negotiation is what firms exist to avoid. Oliver Williamson, who shared the Nobel prize in economics in 2009, argued that markets and firms are rival ways of settling conflicts of interest. When parties are locked together and cannot easily walk away, haggling becomes costly, and a hierarchy that can simply decide does better. The Nobel committee summarised his life’s work as a theory of the firm as a structure for resolving conflict. It also noted the catch: inside a firm, “authority, which mitigates contention, can be abused.”

The budget round is where both halves of that sentence come true at once. The contention is not mitigated; it is merely moved indoors. And the authority that settles it is exercised with little to show for how. The trouble is not that divisions fight over money. Scarcity guarantees they will. It is that the fight is settled by voices—seniority, eloquence, reputation—rather than by a structure that weighs every claim on the same terms. Portfolio management ought to be that structure. Its job is not merely to pick projects, but to resolve the firm’s central conflict in a way the losers can accept.

How the money really moves

Ask executives how they divide capital among divisions and they give the textbook answer. In a survey of more than 1,000 chief executives and finance chiefs, John Graham, Campbell Harvey and Manju Puri of Duke University found that most firms do rank projects by net present value. But they also found that the divisional manager’s reputation and senior management’s gut feel weigh heavily. Finance chiefs, who sit closer to the numbers, were more likely than their bosses to say that internal politics shapes the result.

Theory predicted as much. In a paper published in 2000, David Scharfstein and Jeremy Stein showed how division heads who lobby raise their bargaining power, and how headquarters, itself an agent of shareholders, buys peace with the cheapest currency it has: capital. The upshot is a kind of corporate socialism, in which weak divisions are subsidised by strong ones. The evidence fits. Studying 165 conglomerates, Mr Scharfstein found that divisions in industries with the best prospects invested less than comparable stand-alone firms, while those in the worst invested more. The levelling was most pronounced where managers owned little of the company.

The firm pays twice. It pays once in misallocated capital, and again in the hours its ablest people spend arguing for it. Paul Milgrom and John Roberts gave the second bill a name: influence costs. Neither appears in any budget.

A constitution for the budget round

Williamson’s insight suggests the remedy. If a firm is a structure for resolving conflict, then the budget round needs a better structure, not better speeches. A model of the portfolio can be one. Its virtue is not that it is clever but that it changes what people argue about.

Without it, they argue over outcomes: which division gets how much. With it, they argue over rules, and they do so first. The firm agrees its priorities and how much each matters before anyone knows which projects those priorities will favour. Every proposal is then judged against the same priorities, the same limits and the same scale. The allocation follows from what was agreed. A manager who disputes the result must dispute an input—a weight, an estimate, a limit—openly, where others can test it.

Authority does not vanish; it changes form. A chairman may still insist that his project goes ahead, and sometimes he should. But his insistence becomes a priced exception rather than a quiet adjustment: the board sees what the mandate displaces, in value and in dollars, and the record shows who asked for it. Williamson’s warning about abused authority is answered not by removing authority but by making its use visible.

This is also why losers can accept the result. People will tolerate losing under rules they agreed to before the game began far more readily than losing to a colleague’s charm. Numbers, not voices, settle the allocation. Voices still matter; they are simply heard earlier, when the firm decides what it wants.

A manager who disputes the result must dispute an input, openly, where others can test it.

Voices in disguise

The obvious objection is that numbers can be lobbied too. A manager who cannot win the argument can inflate a project’s value or shave its cost, and a model will faithfully optimise the fiction. Goodhart’s law—that a measure which becomes a target ceases to be a good measure—applies to budgets as much as to central banks. A structure built on numbers can become a place where voices hide.

The objection is fair, but it cuts less deeply than it seems. Inflation is easier to hide in a speech than in an estimate that sits beside every rival’s, on the same scale, in a record that outlives the meeting. One inflated claim may pass; a manager whose claims are always generous leaves a pattern, and outcomes eventually catch up with it. The structure does not make gaming impossible. It makes it more expensive and more visible.

A deeper limit remains. Some conflicts are not about facts but about what the firm should want: growth or margin, the core business or the new one. No model can settle those, and none should try. That is where authority belongs, and it was Williamson’s point that firms exist precisely because someone can decide. The model does not replace the chief executive. It disciplines her, by separating the decisions only she can make from the ones the numbers can.

Argue once

Williamson’s lesson was that institutions exist to economise on conflict. By that test the typical budget round is a failing institution. It is fought every year, settled by stamina and standing, and leaves behind neither a reason the losers can respect nor a record the board can learn from.

The aim is not fewer arguments but better ones, held in the right order. Argue once, and hard, about what the firm wants. Then let a structure everyone agreed to settle who gets what, and make anyone who overrides it pay in public. The firm was invented to stop haggling. In a well-governed one, the loudest voice in the budget round should be the one that set the priorities—and then fell silent.

Numbers, not voices

How the money really moves

A constitution for the budget round

Voices in disguise

Argue once

Further reading