A worker can refuse an offer only for as long as refusal remains possible. The reservation wage sounds like a number carried in the mind: below this, no; above it, yes. But a person does not carry the number alone. He carries rent, dependents, savings, the memory of earlier offers and an estimate of the next one. The offer that was unacceptable a month ago may become acceptable without the work improving or the worker losing a skill. Something else has changed: the cost of continuing to wait. A safety net enters the wage bargain at precisely this point. It gives time a different price. The benefit is paid outside employment, yet it can alter the terms on which employment begins.
John McCall made waiting intelligible as an economic choice. In his simplest search model, offers arrive from a known distribution, one after another, and obtaining another offer has a cost. The worker compares the job in hand with the expected return from continuing. Under these assumptions, the comparison produces a cutoff: accept above it, continue below it. Unemployment can then be the interval in which a person is acquiring costly information, rather than evidence that he attaches no value to work. The model does not make every spell prudent or every rejected offer a mistake. It asks what the searcher could reasonably expect to gain by looking again. The relevant offer, moreover, is the value of the employment opportunity, not a wage figure stripped of hours, duration and conditions.
The logic can be seen in a deliberately small example. Suppose there are only two periods left. A job accepted now pays the same wage in both, and taking it ends the search. Waiting yields a benefit of twenty now, followed by one offer of either sixty or a hundred, each equally likely, for the final period. Assume money is valued linearly, there is no discounting, and the future offer will be accepted. Waiting is worth a hundred in expectation: twenty plus eighty. A current wage of fifty is therefore just sufficient, since it pays a hundred across the two periods. Raise the first-period benefit to sixty and waiting becomes worth a hundred and forty. The cutoff rises to seventy. A wage of sixty, accepted before, is now refused.
The example also shows why a higher cutoff is not a promise that everyone will search longer. An offer of a hundred is accepted under either benefit. The change matters only where it changes the comparison. Nor is the cutoff an immutable declaration of personal worth. If the expected final offer in the example falls from eighty to forty, the cutoff falls by twenty under either benefit. In an actual search, the remaining eligibility period, the arrival of offers, the worker's information and the prospects of the occupation may all change together. McCall's later adaptive model allowed the searcher to learn about the offer distribution. A rule that was sensible under yesterday's information can cease to be sensible today. Insurance supplies room to search; it does not supply the offer for which the worker is searching.
There are two different reasons the room can matter. One is the familiar incentive: income available while unemployed reduces the immediate financial gain from starting a job. The other is the ability to meet expenses while a decision is being made. A worker with savings may already be able to consider several opportunities. A worker without them may have to accept the first feasible one, even if he would prefer to finance a longer search out of later earnings. The same benefit payment changes both the reward for returning and the resources available for waiting. Calling every additional week a distortion confuses the two. Calling every additional week a better search makes the opposite error. The duration records the response; it does not, by itself, identify its cause or its value.
Raj Chetty's analysis gives this distinction a precise form. In his search model, imperfect credit and insurance can make cash especially valuable during unemployment. Additional resources then reduce the pressure to find work immediately, even when they do not reduce the reward attached to accepting a job. A payment that the worker keeps whether or not he returns to work helps separate this resource effect from the incentive built into a payment conditional on remaining unemployed. The distinction is conditional on the model and on what a policy actually changes. A grant paid today and a stream paid over many weeks need not have equivalent effects for someone unable to borrow. Timing belongs in the account, alongside the amount.
Chetty examined differences in benefit responses across households and the durations associated with lump-sum severance payments. The combination supported a substantial role for liquidity in the historical setting he studied. It did not turn severance into a randomized experiment: the paper expressly considered the possibility that recipients differed in other respects. Its lesson for this inquiry is narrower and more durable than a universal percentage. A response to benefits cannot be assigned wholly to reluctance to work merely because it lengthens a spell. Conversely, the presence of a liquidity effect does not prove that the eventual job is more productive or that its wage is higher. It establishes a reason why the worker's preferred timing can differ from the timing imposed by an empty account.
This returns us to the bargain. Suppose, in another illustration, that a match produces a hundred after nonlabor costs, the employer has no alternative return, and the two parties split the surplus above the worker's outside option equally. With an outside option of forty, the wage is seventy: forty plus half the remaining sixty. If the outside option becomes sixty, the wage becomes eighty. The worker's productivity has not changed. The gain from agreement has fallen from sixty to forty. The worker still receives half of that surplus, but his total wage rises because his outside option is higher. The calculation is an assumed bargaining rule, not an estimate of what a benefit system does. A temporary entitlement has to be valued over its actual duration; employers may post wages rather than bargain; and an alternative that merely sounds available cannot perform the work of one the person can use.
A benefit system can therefore affect wages through an outside option, but the individual mechanism is not the whole market result. Firms can change vacancies, requirements or the offers they make. Workers compete for some of the same jobs. A recession can remove opportunities faster than support finances the search for them. The cost of paying benefits also has to be met, and a transfer can alter behavior without creating resources. These are reasons to investigate the equilibrium and the financing, rather than reasons to erase the worker's constraint from the starting point. The relevant comparison is between feasible systems: how much consumption each protects, which searches it permits, which incentives it changes, and what those changes cost.
This is why the question of generosity cannot be settled by counting days alone. A shorter spell may be a successful match or a hurried retreat into work the person soon leaves. A longer spell may be valuable search, a response to a reduced return from working, or both. The distinction asks for evidence about consumption, search, subsequent employment and the alternatives people could actually finance. A system can protect the loss that insurance was created to cover while preserving reasons to take suitable work; how well it does either depends on its design. The argument for protection survives the existence of an incentive response. The argument for attending to incentives survives the existence of need.
The reservation wage, then, is not a minimum written by government into the contract. It is the point at which a person prefers the contract to what comes next without it. A legal wage floor restricts the offers that may be made; insurance changes the situation from which offers are considered. The distinction matters even where both raise the lowest wage a particular worker will receive. One changes the permitted bargain; the other can change the worker's ability to reject it. Neither automatically produces a vacancy, and neither can be assessed without the surrounding market. What insurance can supply directly is a period in which losing the old job need not also mean accepting any new one.
The ability to wait is one of the things a wage bargain prices.
The machine in the next plate can make this period urgent. If the former work disappears, the worker may need time to find a different use for an existing skill or acquire another. No assurance of eventual adjustment pays the present bill. Yet support alone cannot restore the demand that has gone. Insurance belongs between those two facts: an economy can gain from change while the people moving through it bear losses they cannot readily finance. The net beneath search is part of the terms on which that movement takes place. It changes who can afford to wait, and therefore whose judgment has time to enter the price of labor.