The technician arrives knowing the hours of her shift but not which machine will fail during it. The employer knows no more. Their agreement would be of little use if it required a fresh bargain whenever a different repair became necessary; part of its point is to let the day disclose what the work must be. Yet the open space has a shape. A promise to maintain the workshop is not a promise to undertake any act the manager can name, and an agreement to let him choose a repair is not an agreement to let him choose every condition under which she lives. The wage supports a service. The employment bargain also allocates decisions about that service, leaving some choices for later while settling who may make them.
Authority economizes on the need to negotiate each successive instruction. Ronald Coase made that possibility central to his explanation of the firm: obtaining resources through a series of market transactions can itself be costly, and a continuing relationship can substitute direction for some of those transactions. The comparison concerns costs on both sides. A manager can misunderstand the work, waste time supervising it or issue a bad instruction; an independent supplier can sometimes adapt under a continuing contract of her own. There is no theorem that every uncertainty needs an employer. The institutional question is which arrangement lets the relevant decision be made well, at a cost worth paying, after the information needed to make it has arrived.
The agreement can specify a field of discretion even when it cannot profitably specify the final act. Herbert Simon’s formal account distinguishes the action chosen in advance from an accepted set of actions within which the employer chooses later. The worker considers that set and the wage together. Some duties may be fixed, some choices assigned to the employer, and others left to the worker. This division is the substance of the arrangement, not an exception to an otherwise unlimited power. A skilled technician may accept the manager’s choice of which machine needs attention while retaining judgment about how to repair it. The parties have deferred a decision; they have not abolished the need to say whose decision it is.
A broader power of direction can be valuable to the employer while making the work less valuable overall. Consider an invented choice between a routine assignment yielding sixty units of revenue at a cost of ten units of effort, and an extended assignment yielding seventy-six at a cost of thirty. Assume that effort and payment can be compared in common utility-equivalent units, that the wage is thirty-five, and that there are no other costs. The routine assignment gives the employer twenty-five and the worker twenty-five; the extended assignment gives the employer forty-one and the worker five. At the fixed wage, the employer prefers the extension. Joint surplus nevertheless falls from fifty to forty-six. Sixteen extra units of revenue have required twenty extra units of effort. A boundary that keeps the extended assignment outside unilateral direction can prevent that loss.
A right to refuse can also make a worthwhile extension possible on terms both parties prefer. Suppose the extended assignment instead earns ninety. Its joint surplus is sixty, ten more than the routine work, but the worker still bears the additional twenty units of effort. If she may require a new agreement for the extension, an additional payment of twenty-two gives her twenty-seven and leaves the employer thirty-three. Both gain relative to the routine assignment. With revenue of seventy-six, the same proposal would leave the employer nineteen, below his original twenty-five, and he would decline it. Under these stipulated conditions, requiring consent helps separate an extension worth undertaking from one profitable only because its cost can be imposed on the other party. Whether the parties can reach that agreement cheaply is a further question.
The right to refuse and the practical ability to refuse are not the same thing. In the example, accepting the low-value extension leaves the worker five, still more than an outside option of zero. If the only available response is to accept the new task or lose the job, she may accept it even though both parties could have been better off with routine work. If leaving also costs eight, exit yields minus eight and the pressure is greater. These comparisons describe the consequences of available choices; they do not show that the initial agreement authorized the demand. Nor does a worker’s decision to endure a burden prove that it was reasonable to impose. The set of acts someone will perform under pressure can be wider than the set she agreed another person should have power to require.
A legal boundary can therefore operate independently of the worker’s immediate bargaining strength. California Labor Code section 2856 requires substantial compliance with an employer’s directions concerning the service for which the employee is engaged, but excepts obedience that is impossible, unlawful, or would impose new and unreasonable burdens. The wording gives direction a subject and a limit. It does not say that whatever the employee can be induced to tolerate becomes part of the employer’s authority. Applying those terms to a particular instruction requires its facts and the governing law; the institutional point is already visible in the provision itself. An employment relationship can create an obligation to follow directions while also supplying grounds on which a direction must answer to something beyond the manager’s preference.
The means of testing an instruction matter as much as the existence of a boundary. A worker who can ask for an explanation, consult a representative or obtain a decision from someone other than the person issuing the order has a different practical position from a worker whose only recourse is departure. Such arrangements can consume time and can be used badly. They can also bring information into the decision that a command would otherwise suppress. The technician may know that a quick repair will conceal a deeper fault; the manager may know that a different machine is indispensable to an urgent order. A useful division of authority must give each kind of knowledge a route into the choice. Speed alone cannot tell whether the choice was well made.
Custom can make the boundary intelligible, but custom can also preserve a disputed power. A workshop may have a settled understanding that experienced staff choose the method while the supervisor sets the sequence, or that an exceptional shift requires a fresh agreement. Such understandings save the parties from renegotiating every detail. They also need a way to change when work changes. A practice endured by successive workers may reflect consent, necessity or a failure to find a safe occasion to object. Simon’s account makes room for informal and collective terms; it does not certify every inherited practice as an agreement advantageous to all. The question is what the practice authorizes, how it became authoritative, and whether those subject to it can contest its extension.
The economic reason to leave work unspecified does not settle the justice of the resulting hierarchy. A manager may need room to redirect a team when circumstances change, yet the scope of that room, the burdens it distributes and the procedures that restrain it remain questions for the institution. The wage cannot answer them merely by being high enough to attract someone. Compensation can alter the division of a gain; it cannot make an impossible instruction feasible or give an unlawful act a lawful character. Conversely, a limit on authority need not defeat cooperation. It can define the stable conditions within which both sides are prepared to invest, learn and rely on decisions made by the other. A bounded power may be more useful precisely because its boundary can be trusted.
The bargain can leave the work open while placing a boundary around the power to choose it.
An employer buys neither a finished list of acts nor a person without a remaining will. He obtains a service under an arrangement that assigns some future decisions to him, and the value of that arrangement depends on how its authority is bounded and exercised. The worker’s power to leave is one part of the account; the terms she may insist upon while remaining are another. Read in this way, employment reveals the contract’s peculiar achievement. Words can organize decisions they do not themselves make, allowing people to cooperate across a future neither can describe in full. The final plate asks what it means for those words to bind, when performance remains an act that another person must still choose. The argument remains open.