The Question Nobody Had Asked

Ronald Coase was twenty-one and travelling around America on a scholarship, visiting factories, when he noticed something that should have been obvious and was not.

Economics taught that the price system allocates resources efficiently. Prices carry information, people respond to them, and the result is a coordinated economy nobody designed. Excellent.

Inside a factory, none of this happens. Nobody bids for the machine. No price is negotiated when one department passes work to another. A manager says do this, and it gets done. Vast quantities of economic activity (most of it, in a developed economy) are organised by instruction rather than by price.

So why does the market stop at the factory door?

His answer, published in 1937, is that using the price system has costs of its own. Finding a supplier. Comparing them. Negotiating terms. Writing a contract that anticipates the contingencies. Monitoring whether the work was done. Enforcing it if it was not. These are transaction costs, and where they exceed the cost of simply telling somebody to do it, an organisation appears.

A firm, on this account, is an island of planning in a sea of market exchange, and its size is set by where the two costs cross.

This also answers the reverse question, which is why firms do not simply keep growing until one of them is the whole economy. Each activity brought inside is one more thing to monitor, one more layer for information to pass through and degrade. The returns to management diminish. The firm expands until the cost of organising one more transaction internally equals the cost of buying it on the market, and stops.

Oliver Williamson developed the most empirically successful branch of this, and it turns on one specific problem.

Suppose you are a supplier and a car manufacturer asks you to build a press that stamps their particular body panels. The press costs millions and, once built, is worth a great deal to that manufacturer and almost nothing to anyone else. The investment is asset-specific.

Now you are exposed. Once the press exists, the manufacturer can reopen the price negotiation, and your alternative is scrap value. This is the hold-up problem, and the important thing about it is what happens before anybody behaves badly: anticipating the exposure, you either refuse to build the press or demand a price that reflects the risk, and a valuable investment does not happen.

Unifying ownership solves it, because you cannot hold yourself up. Williamson's prediction, that vertical integration should appear where asset specificity is high, has been tested extensively and holds up well, and he shared a Nobel for it.

There is an observation lurking here that people on both sides of a long political argument tend to avoid.

A firm the size of Walmart or Amazon allocates resources on the scale of a mid-sized national economy, internally, by plan. Budgets, targets, directives, internal logistics, not prices, not bidding, not markets. If central planning were impossible in principle, these organisations could not exist.

What the evidence actually shows is subtler and more useful: planning works, and its costs rise with scale, complexity and the amount of local knowledge involved. That is precisely Coase's boundary condition, and it explains both why enormous firms exist and why they are not larger.

The everyday version of all this is the outsourcing decision, and the reversals are more informative than the original moves.

A company compares the hourly rate of an outsourced cleaning contract with the cost of employing cleaners, finds the contract cheaper, and outsources. Some years later it brings the function back in-house.

What was missing from the original comparison is everything Coase named. Writing a specification for what "clean" means. Inspecting against it. Handling the dispute when standards slip. Losing the ability to redirect someone to a spillage in the middle of a shift, because that is not in the contract. None of these appear in the hourly rate, all of them are real, and they are systematically underestimated because the comparison that motivates the decision does not have a column for them.

That is not an argument against outsourcing. It is an argument that the boundary of the firm is set by a cost most organisations do not measure, which is why it keeps moving back and forth.