Essay 01

The planned economy in every boardroom

Firms allocate more capital than the markets do, and they do it without prices. The tool that could supply them was invented for Soviet planners.

In 1937 Ronald Coase asked an awkward question. If markets allocate resources so well, why do firms exist at all? Inside a company, he observed, the price mechanism is switched off. A manager does not bid for an engineer; he is assigned one. Coase borrowed Dennis Robertson's description of the result: firms are “islands of conscious power” in an ocean of market co-operation.

The islands have grown. In 2012 McKinsey estimated that America's multi-business companies allocated or reallocated around $640bn of capital a year—more than the equity and corporate-debt markets raised combined. How they do it is less impressive than how much they do. In a study of 1,616 American firms between 1990 and 2005, the capital a third of business units received each year correlated with what they had received the year before at 0.99. The economy's biggest allocators of capital mostly copy last year's answer.

The usual diagnosis is that firms lack good ideas, or the nerve to back them. The better one is that they lack prices. Portfolio management is a competition for scarce resources inside the firm—money, people, capacity—and that makes it a problem of economics, not of scoring. An economy without a market needs some other way to discover what things are worth. Mathematical optimisation supplies one. It proposes an allocation and, as a by-product, reports the consequence of every limit placed upon it.

Anchored to last year

Start with the inertia. In one experiment McKinsey's researchers asked executives to divide capital among the units of a fictitious company. Those given only the market outlook allocated according to prospects. Those also shown the previous year's budget clustered around it, though it had little to do with future returns. Nobody's career was at stake; the anchor held anyway.

Inertia, in other words, is not laziness. It is what happens when there is nothing better to anchor on. A budget round without prices is settled by precedent and by whoever argues most forcefully. The cost is measurable. The third of firms that moved resources most—shifting, on average, 56% of their capital between units over 15 years—earned total shareholder returns about 30% higher than the least active third.

Lists are not portfolios

Why do firms find it so hard to move? Partly because the standard tools answer the wrong question. Most portfolio processes score projects, rank them and fund down the list until the money runs out. Even in the simplest case—independent projects, money the only constraint—that fails, because projects cannot be split: funding down a ranking routinely strands money that a different combination would have used. In practice projects are rarely independent, and money is rarely the only thing in short supply.

Scarcity is a property of combinations, not of lists. Two highly ranked projects may need the same dozen data engineers; a modest one may be the only use for capacity that would otherwise sit idle. A ranking cannot say what one more dollar would buy, what one fewer engineer would cost, or what the ring-fence around a favoured division is worth. It answers “which project is best?” when the real question is “which set is best, given everything we are short of?”

The planner's discovery

The answer to that question came from an unlikely place. In 1938 Leonid Kantorovich, a young Leningrad mathematician, was asked by the Laboratory of the Plywood Trust how to get the most output from a set of machines and materials. He saw that the problem was a mathematical one—maximising a function subject to many constraints—and in 1939 published a method for solving it.

The method had a curious by-product. Alongside the best plan it produced a number for each scarce input: what an extra unit of it would be worth at the margin. Kantorovich called these “objectively determined valuations”, a phrase chosen partly to disarm Marxist critics. Everyone else would recognise them as prices. A tool built to run production without markets had rediscovered the market's central signal. It earned Kantorovich a share of the Nobel prize in economics in 1975, with Tjalling Koopmans, for work on the optimal allocation of resources.

The irony is that firms—which are, by Coase's account, economies without markets—have made little use of the idea where it matters most. They apply optimisation readily to refineries, rosters and delivery routes. They rarely apply it to the portfolio, the one place where the whole firm competes for the same scarce pool.

What the consequences reveal

Every limit a firm places on its portfolio has a consequence. Some limits are hard: the budget, the law, the number of qualified people who exist. Many are choices: a mandate that a chairman's project must proceed, a rule that each division keeps its share, a cap on headcount set before anyone knew what the work would be. Some of these choices are wise. Others survive because nobody can see what they cost.

Once consequences are visible, the conversation changes. A board that learns that protecting one division's share forgoes a fifth of the portfolio's strategic value, or that its binding constraint is not money but a single team, argues about different things. The question moves from “whose project?” to “is this rule worth what it costs?” The consequence is measured first in strategic value, but it can be translated into the dollars it would take to buy that value back—a unit boards understand far better than scores. Because projects are all-or-nothing, it is found by comparing the portfolio with the limit and without it, not read off a tidy marginal price. Nor is the consequence always a cost: relaxing the right limit can reveal value the firm did not know it was forgoing.

Most boards approve a portfolio without knowing what their own rules cost.

What the model cannot know

The method has limits, and most lie in what it is fed. A model is only as good as the values and estimates put into it, and those are judgments, often made by people with a stake in the outcome. Optimisation cannot make a biased estimate honest; it can only show how much the answer depends on it. Nor does it say what the firm should value. Given the priorities, it finds the best set; setting the priorities remains the board's job.

Projects are also lumpy. A firm cannot fund 60% of a factory, so the consequences of a limit jump rather than glide, and a small change in an estimate can flip a large decision. That is why an optimum is a proposal, not a verdict. Timing matters too: McKinsey found that over periods of under three years, heavy reallocators did worse than steadier peers. The pay-off came later.

The evidence, too, has limits: it shows that inertia is costly, not yet that optimisation cures it. Still, none of this rescues the ranked list. Optimisation does not abolish judgment; it relocates it—away from arguing over the answer and towards arguing over the assumptions, which is where the argument belonged all along.

The visible hand

Coase's islands of conscious power have grown into continents, yet they still steer largely by precedent. Optimisation would not have saved Soviet planning. The point is narrower: mathematics born in a command economy can expose the costs that corporate planning leaves hidden. The companies that pull ahead are unlikely to be those with the best projects. They will be those that know what each of their own rules is costing them. In the market, the hand that allocates is invisible. Inside the firm it is visible already; it has simply never been asked to show its consequences.