The carpenter who agrees to deliver a table has promised a result. The carpenter who takes a job in a workshop may spend Monday repairing a cabinet and Tuesday making a door, although neither task appeared in the agreement by which she entered. Between the wage and the work lies a decision someone has yet to make. The employer has bought a claim upon that decision: within a range the relationship permits, he may select what she will do after he learns what needs doing. Her labor is the service to be performed; the distinctive bargain concerns who will choose its particular use. The worker has committed herself to a bounded authority whose instructions will arrive later.
The interval between agreement and instruction is what makes authority useful. In Herbert Simon’s account, the parties can settle the wage and the range of acceptable actions before they can profitably settle the action itself. A customer may change an order; a machine may fail; the task most valuable on Tuesday may be unknowable on Friday. Specifying a single performance in advance can sacrifice the value of responding to what happens. Simon calls the agreed range an area of acceptance. Inside it, the employer chooses the action; outside it, the original bargain supplies no such permission. The range matters as much as the wage, because the same payment can buy an agreeable afternoon or a succession of instructions the worker would have priced quite differently had she known their content.
A small workshop can reveal the value of leaving the choice open. Take the constructed example developed in the earlier Employee. Two equally likely kinds of order may arrive. Task A earns twelve units on the first and four on the second; task B earns four on the first and twelve on the second. Either task costs the worker two units of effort, measured in the same units as payment, and neither requires any other input. A commitment to A before the order arrives earns eight on average and leaves six after effort; a commitment to B does the same. Permission to choose between them after the order arrives earns twelve and leaves ten. At a wage of five, the employer receives seven and the worker three. Four units of joint gain have come from waiting to decide. Nothing in the gain required the employer to become wiser than the worker; it required the choice to be made when the relevant information was available.
The gain belongs to adaptability before it belongs to any particular form of command. If the two orders can be described in advance, verified cheaply and written into an enforceable agreement, the parties can contract for A when the first arrives and B when the second arrives. That contingent bargain also produces ten. They might instead let the worker select the task, if she sees the order and has reason to respond appropriately. Authority becomes attractive when leaving a choice to the employer is less costly than specifying, verifying or renegotiating the required response. The workshop establishes what delayed choice can be worth. To establish why a boss should hold that choice, one must also ask who knows, who can prove what happened, and whose incentives will make the valuable response the one actually chosen.
The employer’s freedom to choose can also create the loss the agreement must prevent. Add a third task, C, which earns thirteen under either order but costs the worker eight. At the same wage of five, the employer receives eight by ordering C, against seven from the appropriate A or B. He prefers C. The worker receives minus three, and their combined gain falls from ten to five. The extra unit entering the till has cost six additional units of effort. A choice profitable to the person issuing the instruction has destroyed half the available surplus. This is why the boundary of the acceptance area cannot be treated as a polite qualification attached to a theory of unlimited obedience. It is part of the economic arrangement that determines whether authority will produce its promised gain.
A higher wage can make an onerous command acceptable without making it economical. Pay nine for the enlarged range and the worker obtains one from C while the employer obtains four. Both can prefer that arrangement to an outside option of zero, yet together they still receive only five. Their willingness to sign has not recovered the five units destroyed by the badly allocated choice. Nor can a fixed increase in wages, once agreed, change the employer’s preference between tasks whose revenues differ: the same wage is subtracted from each. The parties need a different range, a different allocation of discretion or a payment rule that changes the relevant incentive. The terms of consent determine who bears a cost; they do not by themselves show that the cost was worth incurring.
The wage pays for work; the employment bargain also assigns the choice of what that work will be.
An employment agreement can divide discretion instead of placing it all in one pair of hands. Simon’s account allows particular duties to be fixed, some decisions to be assigned to the employer and others to remain with the worker. A workshop may specify the hours, permit the manager to select the customer’s order, and leave the craftsperson to decide how to cut the joint. Where the parties’ interests sharply diverge, a fixed condition can protect the bargain; where future needs are uncertain and their interests nearly coincide, discretion can save repeated negotiation. Skill may also place the useful information with the worker. The arrangement then has to preserve room for judgment rather than reduce every movement to an instruction from someone who knows less about the work.
The possibility of leaving a job gives the bargain a limit, but the cost of leaving affects how that limit works. In the workshop, an outside option of zero was assumed for convenience. Give the worker a better alternative and she can reject more demanding terms; make departure costly and a previously unattractive instruction may become the least costly available choice. Neither situation makes the worker the employer’s property. It changes the pressure under which she exercises her remaining choice. The economic value of adaptability and the distribution of power within employment therefore require separate judgments. An arrangement can coordinate production effectively while leaving its burdens badly distributed; a worker’s agreement to enter does not tell us how costly it will later be to object.
The word obedience conceals the particulars on which its value and its danger depend. Obedience to which instructions, during which hours, using whose judgment, with what opportunity to challenge a demand? These questions specify the thing exchanged. They also distinguish the power to organize an activity from a general entitlement to rule the person who performs it. A wage is visible and readily compared. The surrendered choices may be harder to enumerate, especially when the very purpose of employment is to leave some future acts unnamed. Reading the agreement as an allocation of discretion makes those choices visible without pretending that all of them could have been settled in advance.
The worker has sold a service under an agreed arrangement for selecting its use. That arrangement may enlarge what both parties can achieve, and it may give one of them opportunities to advance his own interest at the other’s expense. The range of authority, its reward and its limits belong in the same account. A firm can consequently be constituted by agreement and still contain a hierarchy: the agreement can be precisely what assigns the hierarchy its decisions. The next plate follows that thought into the contracts surrounding the firm, to ask whether the island disappears when its foundations are named, or whether naming them tells us how the island is built.