If the firm exists because directing resources by command can be cheaper than buying them in the market, an awkward question follows at once. Why not, then, direct everything by command? Why not gather the whole economy into a single great firm and dispense with the market altogether — the very proposal that, under the name of central planning, the twentieth century tried and abandoned at terrible cost? The answer must be that command has costs of its own, costs that grow with the size of the thing commanded; and the firm's true size is fixed not by how much the market wastes but by where the waste of the market and the waste of the office come, at the last bale and the last memorandum, to exactly balance.
The shape of the answer was already in Coase, and it is the shape of every answer in this work's sixth volume: a margin. A firm should bring an activity in-house, Coase reasoned, just so long as doing so costs less than buying the same activity in the market; and it should stop — leave the next activity to be bought — at the point where the two costs meet. The boundary of the firm is that meeting point, the make-or-buy line traced out transaction by transaction. It is an elegant principle and, for a long time, a frustrating one, because it named the criterion without supplying it. To say that a firm integrates whatever is cheaper to organize within tells us nothing until we can say what makes organizing within cheaper — and lacking that, the theory could only point at any boundary already drawn and pronounce it efficient, having no way to predict where the line would fall before it fell. The puzzle of the boundary was not why it existed but what governed it; and that took another economist, and another generation.
The economist was Oliver Williamson, and the answer he gave turned on a single, fertile idea: that what pulls an activity inside the firm is not its cost or its complexity but a particular kind of vulnerability. Most things a firm needs it can buy from anyone — bolts, electricity, haulage, the standard services of a competitive market, where if one supplier disappoints, another stands ready. But some things require an investment shaped to one partner and almost worthless away from him: a factory built beside a single buyer's works and tooled to his specifications; a component designed for one model and no other; the deep, particular knowledge a worker accumulates of one firm's machines and ways. Williamson called the property asset specificity, and saw that it changes everything about the safety of dealing in the market. So long as what you buy is generic, the market protects you, for you can always turn elsewhere. The moment what you need is specific — built for this relationship and stranded outside it — the protection falls away, and a hazard opens that the market cannot close.
The hazard is the holdup, and it is worth seeing exactly. Imagine a parts-maker who, to win a carmaker's business, must build a plant useful only for that carmaker's panels. Before he builds, he is one bidder among many, and the carmaker must court him; the instant he builds, the courtship ends, for his plant is now worth a fortune to this one buyer and scrap to all others. He is, in Williamson's phrase, fundamentally transformed — from a party with options into a hostage — and the carmaker, knowing it, may press for a lower price, daring him to refuse and eat his sunk cost. The parts-maker foresees all this, and so may decline to build the efficient plant at all, or build it only behind a thicket of costly contractual armor. The clean solution is to dissolve the two parties' separateness: let the carmaker own the plant. Under one roof there is no one to be held up and no one to do the holding, and the quarrel the market would have settled by extortion is settled instead by an order from above. Where assets are specific, the firm swallows the market — and that is what vertical integration is.
This explains why firms reach inward, but not why they ever stop — and the stopping is the boundary's other half, and the answer to the question this plate began with. If integration cures the holdup, why does the carmaker not also make its own steel, grow its own rubber, drill its own oil, until one firm has absorbed the world? Because authority has rising costs of its own, and they catch up. The manager who would direct everything cannot know everything; his bounded attention, stretched too far, governs each part worse than a focused rival would. Incentives within the firm are duller than the market's — the salaried division has less reason to strain than the independent supplier whose survival rides on each sale. And the great organization breeds its own waste, the empire-building and budget-fighting that consume effort better spent on the work. Marshall had marked the two faces of this long before, the economies a firm wins from its own organization set against those it draws from the industry around it; the boundary is the seam between them, and falls where the office's mounting costs overtake the market's.
Two things follow, and the first is that the boundary is not a fixed feature of an industry but a standing decision, taken and retaken. Make or buy is a question a firm answers afresh for every component and every service, and its answer changes as the conditions change. When the costs of transacting are high — communication slow, suppliers few and far, contracts hard to enforce — the safe course is to make, and firms grow great and self-contained, doing everything within. When those costs fall — when networks are cheap, supplier markets deep and competitive, contracts easily written and policed — the safe course shifts toward buy, and firms slim to what they do best and purchase the rest. The long movement toward outsourcing in our own time, the breaking of the vast integrated firm into a core ringed by contractors and suppliers, is not some change in the nature of business but the boundary sliding outward as the cost of using the market has fallen. The line is alive, and it moves.
Honesty requires the theory's rivals and soft spots be set down too. Williamson's account, for all its reach, is not the last word. A powerful alternative, the property-rights theory of Grossman, Hart, and Moore, accepts that incomplete contracts are the root of the matter but denies that integration is a costless cure: to absorb a partner is to strip him of the residual right to decide, weakening his incentive to invest even as it strengthens one's own. Ownership, on this view, is two-edged, to be granted to whichever party's effort matters more, and the boundary balances incentives on both sides rather than merely fleeing the holdup — a refinement the plate on the asset takes up. Others, recalling the previous plate's quarrel, doubt there is any sharp line between firm and market at all, seeing instead a continuous spectrum, from the one-off purchase through the long contract and the franchise to full ownership, with no natural break. And the very episodes most often told to illustrate the holdup — the carmaker and its body-supplier above all — have been picked apart by historians until it is no longer clear they happened as the parable says. The mechanism is real; its dominion is contested.
Set these disputes aside and the boundary still teaches the lesson this work has pursued from the start. A price, we have held, is a verdict on a hidden contest, mistaken for a fact of nature; and the firm's boundary is the line that decides where such verdicts are handed down by the market and where, instead, they are reserved to the quiet authority within. To buy a thing is to submit it to the market's pricing; to make it is to withdraw it from that pricing and settle its fate by command. Which of the two shall govern any given exchange is not dictated by the nature of the thing but chosen, on a reckoning of costs and hazards — of who could hold up whom, of where the open deal is too dangerous to risk. The boundary of the firm is thus the frontier of the priced world, the limit of the market's writ; and like every frontier it was drawn by someone, in the face of some contest, and could have been drawn elsewhere.
And this should change how we read an industry's very shape. We are inclined to take the arrangement of an industry — which firms, of which sizes, making what and buying what — as a kind of natural fact, the form the business was bound to assume. It is nothing of the kind. Every boundary in it is the residue of a choice that might have gone otherwise: the carmaker that buys its bodies might have built them, the studio that owns its lots might have rented them, the grocer that runs its own delivery might have hired it out. What looks like the anatomy of an industry is the frozen accumulation of countless make-or-buy decisions, each one a small judgment about how far the price system should be let in. The price of the finished thing — the car, the film, the delivered parcel — carries the whole of that hidden architecture within it: not the verdict of a single open market but the sum of a hundred prior verdicts about where markets should be allowed to reach, and where command should rule instead.
To buy a thing is to trust it to the market; to make it is to withdraw it from the market and rule it by command — and which shall govern is chosen, never given.
So the two opening plates of this volume close a single arc. The first asked why the islands of planning exist at all, and found the answer in the cost of using the market; this one asked how far each island spreads, and found the answer in the cost of leaving it — in the asset too specific to trust to a stranger, and the office too large to govern well. Between them they dissolve the comfortable picture of an economy that is simply, naturally, a market. A price is a verdict on a hidden contest, mistaken for a fact of nature — and the firm shows that even whether a thing is to be priced at all, or made within the silent precincts of an organization, is itself such a verdict, rendered at the boundary, revisable, and never written in nature. The line between the market and the firm is the line between the priced and the planned; it runs through every industry; and it was drawn, as all such lines are drawn, by choice contending with cost. The argument remains open.