A large firm must distribute decisions if it is to remain capable of making them. The manager who understands one product's customers cannot also know every production difficulty in a second business and every investment opportunity in a third. The multidivisional arrangement assigns operating responsibility to units organized around businesses, products, or territories, while a central office compares their performance and decides how much of the enterprise's resources each should receive. Production and selling can be joined within a division, close to the activity they serve; choices among divisions remain above it. The firm gains several centers of practical judgment without surrendering every common decision. This is the organizational problem that follows the charter: a legal person can hold the whole enterprise together, while the knowledge needed to direct it remains scattered among many actual people.
A division's operating autonomy is therefore joined to a dependence on the capital allotted to it. Its manager can propose a new plant, revise a product, or explain a disappointing year, while headquarters decides whether the proposal deserves resources that another division also wants. The comparison creates an internal capital market. The word market identifies a contest among uses of funds, though the decision is made through an authority that can reject a request and redirect the money. Cash earned in one business need not return automatically to that business. A division with reliable earnings may have few valuable new investments; another may have opportunities whose promise has yet to become cash. Common control makes their opportunities available for comparison within one budget, subject to the enterprise's existing obligations and the funds actually available to it.1
Better allocation can add value even when the company obtains no additional finance. Suppose two divisions, A and B, share a budget of two hundred. Each can use one or two successive blocks of a hundred, with all receipts expressed at a common valuation date. A's first block yields 150 and its second adds 140; B's first yields 120 and its second adds 110. Giving each division one block produces 270, or seventy above the outlay. Giving both blocks to A produces 290, or ninety above the same outlay. The improvement is twenty. B's first investment remains worthwhile in isolation: it yields twenty more than it costs. Under the shared constraint, however, it displaces an investment in A yielding forty more than its cost. The head office's useful judgment concerns the opportunity given up by each allocation, including the good opportunity that a better one leaves unfunded.2
The authority to reallocate matters because the manager losing a budget may have reason to resist the company's gain. A manager can value a larger operation, more staff, or the chance to carry a favored project through. Headquarters, controlling several projects, can benefit from an improvement elsewhere that gives this manager little satisfaction. Stein's model makes that difference precise: a supervisor's interest can extend across the competing projects even while the supervisor remains self-interested. The benefit is conditional on incentives to learn about their prospects and to select a valuable combination. Interposing another layer of management also carries costs, including weaker incentives for those whose control is reduced. In the numerical example, an extra organizational cost of twenty-five would more than consume the allocation gain of twenty. The additional office earns its place only through gains large enough to meet the costs of having it.3
The center's advantage depends on what its comparisons can reveal. Suppose two proposals' true net values are forty and twenty. Overestimating each by ten preserves their ranking; overestimating only the second by thirty reverses it. Related activities may be easier to compare because some errors affect them together. Stein develops that possibility while distinguishing correlated errors in appraisal from correlated business outcomes. The distinction matters. If two divisions always have exactly the same opportunities, there may be little benefit from shifting a fixed budget between them. If their opportunities differ but their accounts can be understood on a common basis, comparison may be useful. A shared product label proves neither condition. The central office must know enough about the businesses to distinguish a promising use of capital from a persuasive description of one.4
Delegation also creates a problem of how much the center should try to know. A supervisor close to a small group of operations may understand them better than a distant head office, yet favor that group when the whole enterprise would gain from moving resources elsewhere. Additional layers can improve the collection of knowledge while giving it more local interests to pass through. Stein presents this as a further organizational tradeoff, rather than a conclusion that either a flat or a layered hierarchy must prevail. The accounts have to support that inquiry. A division's current profit records what it has earned under existing arrangements; an investment decision asks what another commitment would yield. The two can diverge even with honest reporting, as the cash-producing business with few new opportunities shows. Review must connect measured performance to the decisions the measure is meant to guide.5
The division makes a large enterprise manageable by assigning some judgments locally and reserving others for the whole. Its strength rests on that allocation of responsibility: enough discretion to use particular knowledge, enough comparison to challenge a local claim, and enough authority to carry the resulting decision through. Each part can then act with purpose while remaining answerable for resources that have uses elsewhere. The arrangement still leaves open who should ultimately appoint the people conducting this comparison and receive the surplus they preserve. Investors commonly occupy that position; workers and customers can occupy it too. The next plate, The Cooperative, changes the owning constituency and asks how the work of governing changes with it.