A customer orders a table; an employer hires a carpenter for the week. The first bargain can name the result before the work begins. The second can leave Tuesday's task undecided until Tuesday brings its information. The wage purchases a particular kind of flexibility: within an accepted range, one party can select what the other will do without negotiating a fresh price for every instruction. That range is part of the bargain. A promise to work does not mean that every conceivable command has been purchased, or that the worker's own judgment has been surrendered. The employment relation draws a boundary around decisions that will be made later, then places an authority within it.
Herbert Simon made this delayed choice the center of his formal account. Instead of contracting now for one action, the parties can agree now to a set of acceptable actions and let the employer choose from that set after learning more. The worker's willingness to accept the arrangement depends on the wage and on what the permitted actions may cost her. The employer values the ability to wait. There can be a gain from leaving the action open, but a larger set is not automatically a better bargain for both. The model asks which decisions should remain fixed, which should be delegated to the employer, and which can remain with the worker. It gives the area of authority an economic explanation precisely by treating its limits as something that requires explanation.1
Give the workshop two possible tasks, A and B, and two equally likely states of demand. In the first state A yields 12 units of revenue and B yields 4; in the second, the figures reverse. Either task costs the worker 2 units of effort, measured on the same scale. Choosing a task before demand is known yields expected revenue of 8 and a joint surplus of 6. Choosing the appropriate task afterward yields revenue of 12 and a surplus of 10. With a wage of 5 and no other costs, the flexible arrangement leaves 7 for the employer and 3 for the worker. The wage divides the surplus; the ability to adapt creates the additional four units. These are stipulated numbers in a small example, not a measurement of the productivity of employment.2
Now add a third task, C. It yields the employer 13 in either state but costs the worker 8 in effort. At a fixed wage, an employer attending only to revenue prefers C to the task that yields 12. Joint surplus falls to 5. A worker offered the original wage of 5 and expecting C would receive minus 3, so the old payment no longer makes the widened assignment attractive. Raising the wage to 9 leaves the worker 1 and the employer 4; both receive less than they did with the narrower choice between A and B. No rearrangement of the same five units can give them the ten units the better assignment produced. A boundary on authority can therefore preserve value, as well as distribute protection. The right to issue another instruction is not always a productive asset.3
The comparison still owes an account of its alternative. If demand can be specified clearly and its state can be verified at acceptable cost, the parties might write a contingent contract: perform A in the first state and B in the second. That contract obtains the same ten units without leaving the choice to a later discretionary order. In a real comparison, writing the conditions, observing them, communicating a change, and settling a disputed instruction all consume resources. So does supervising an employee. The value of waiting for information does not establish that authority is always the cheapest way to wait. It shows what an employment agreement can accomplish when listing and enforcing the relevant contingencies would be more burdensome than defining a usable field of discretion.4
Nor need the employer be the party who learns most. A carpenter standing beside the timber may see a defect that the office cannot see. Giving her discretion can improve the choice, while an order from someone less informed can spoil it. The relevant arrangement may accordingly reserve some decisions to the employer and leave others to the person doing the work. Its boundaries can be expressed in a written agreement, a negotiated practice, or the protections governing the relationship. Their economic function is intelligible without pretending that every workplace has the same boundaries or that consent makes every exercise of authority just. A useful instruction and an oppressive one can arrive through the same channel. How instructions are challenged, revised, and constrained belongs to the institution the wage helps sustain.5
The worker agrees to a range of future actions because an agreement with room to adapt can make a continuing relationship worth entering. The acceptable wage and the acceptable range have to be considered together. A more demanding range can require compensation, a narrower range can improve decisions, and some instructions can remain outside the bargain altogether. This is the small constitution inside the familiar exchange of work for pay. It governs the passage from a contract made yesterday to a decision made today. Yet the person giving the instruction may herself be employed. Above the workshop manager can stand directors, and behind the directors owners who do not perform the daily work. Once authority has been delegated inside the firm, the next question is how the shareholder governs those to whom its exercise has been entrusted.