The farmer has a crop to sell, and the processor has a price to offer. There may be many farmers and only one plant within reach of the harvest before it spoils; the truck can travel, but the crop cannot wait indefinitely for a better destination. The processor’s advantage begins in that unequal patience. He can refuse some of what is offered and thereby make the rest cheaper to acquire, just as the single seller can withhold some of what he owns and make the rest dearer to buy. The mirror is exact in its direction and unsettling in its appearance. From the processor’s side of the invoice, the exercise of power looks like a saving.
A buyer has this power when obtaining more requires him to offer better terms, and those better terms also reach purchases he would otherwise have made more cheaply. Imagine a common price paid for a crop of uniform grade. The last load costs its own price plus the increase paid on the preceding loads. His marginal expenditure therefore exceeds the price the farmer receives. To decide whether to buy another load, the processor compares its value to that larger expenditure. He stops while the next load would still be worth more to him than the price needed to induce its production. A trade with a gain available on both sides has been abandoned because completing it would make his other purchases more expensive.
The difference can be made visible without pretending to measure a real industry. In an invented market, let the price required to bring forth q loads be 20 + q, and let the value to the processor of the next load be 80 − q. Competitive buying would bring the two together at thirty loads and a price of fifty. A single buyer paying one price instead equates the next load’s value to 20 + 2q, his marginal expenditure. He buys twenty, pays forty, and values the twentieth load at sixty. The farmers receive less for a smaller crop. The processor gains by stopping, even though there are ten further loads whose value would exceed their cost of supply.
The buyer’s gain is only part of what the restriction does. With thirty loads, the total gain from trade is nine hundred: four hundred and fifty on each side. With twenty, the total is eight hundred. The suppliers retain two hundred and the buyer six hundred. His gain of a hundred and fifty accompanies their loss of two hundred and fifty; the remaining hundred belongs to nobody. It is the value of the ten purchases that would have been worth making. This separates a dispute about who receives the surplus from the disappearance of surplus itself. A low invoice can conceal both, and a comparison confined to the buyer’s expenditure will see neither clearly.
The labor market gives this mechanism its most consequential setting because the thing withheld is a chance to work. A common wage makes the next hire expensive in the same way as the next load: an employer may have to improve the offer for existing workers as well as for the recruit. Yet the isolated company town is only one route to an upward-sloping supply of labor to a firm. Search takes time; a vacancy must be found and an offer obtained; a job differs from another in its hours, location, colleagues and demands. An employer can face a finite response to its wage even among numerous other employers. Card’s account of modern monopsony brings search frictions and differentiated jobs into the same field of view. The important object is the worker’s accessible alternative, which a count of firms cannot by itself describe.
The familiar gap between wage and product needs a carefully named product. In the elementary employer’s problem, it is the marginal revenue product: the additional revenue generated by another unit of labor. Where the employer also has power over the price of its output, that revenue differs from the selling price multiplied by the extra physical output. Nor does an accounting average of revenue per worker reveal either marginal quantity. Even after the appropriate value has been found, the denominator matters. In the crop example, a value of sixty and a payment of forty make the shortfall one-half of the payment, but one-third of the value. Both fractions describe the same twenty. Treating them as interchangeable manufactures a disagreement where the underlying transaction has not changed.
A fall in the price paid does not, by itself, establish that this power has increased. Suppose instead that an improvement in production lowers the crop’s supply schedule from 20 + q to 10 + q while buying remains competitive. The price falls from fifty to forty-five, purchases rise from thirty to thirty-five, and total gain from trade rises to twelve hundred and twenty-five. This saving comes from a larger set of worthwhile trades. The monopsony saving came from declining some of them. An investigator must therefore ask what changed in costs, opportunities and quantities before reading the buyer’s power from a falling price. Cheapness is an outcome shared by mechanisms with very different consequences.
Bargaining over a common gain also differs from suppressing purchases to cheapen them. A large retailer and a large manufacturer may negotiate a payment that shifts the division of their joint earnings while leaving the efficient quantity unchanged. A contract may separate a fixed payment from the price of another unit. At the limiting extreme, a buyer who knows every supplier’s reservation price and can pay each separately need not raise earlier payments to obtain the next unit. In the invented crop market, such a buyer could acquire all thirty worthwhile loads and leave the sellers almost none of the gain. The allocative loss would disappear while the distribution became still harsher. Quantity and division must be examined together; neither stands in for the other.
A cheaper purchase can conceal a smaller gain from trade.
Countervailing institutions change these margins rather than merely expressing disapproval of them. A floor under the common payment can, over a range, remove the buyer’s incentive to suppress purchases in order to lower the price on units already bought. A cooperative may bargain for its members; improved transport may make another outlet usable; a worker may gain a credible alternative when search or switching becomes less costly. Each intervention changes a different part of the problem. Its effect depends on how terms are set and how buying responds. The single-buyer model explains why greater pay and greater employment can sometimes be obtained together. It supplies no universal schedule by which every increase in a floor or in bargaining strength produces more of both.
The power of the buyer lies in the terms he can worsen without losing enough supply to make the worsening unprofitable. That definition reaches beyond the lone plant, but it does not turn every hard bargain into the same offense against production. The question is what opportunities the sellers can actually take, how payment responds to the quantity bought, and what the buyer gives up when he refuses another unit. Once there are several large buyers or sellers, those choices begin to turn upon one another. The mirror has shown how one party can move a price by moving a quantity. The next plate asks what happens when the other parties can move it back.