Before two parties can trade, they must find each other, and finding is costly. The buyer who would treasure a thing and the seller who would gladly part with it may live a street apart and never meet. Between them stands neither a disagreement over price nor a shortage of willingness, but an ignorance that leaves both poorer. The matchmaker earns his fee by crossing that distance. He knows which house and which buyer, which worker and which post; he gathers the scattered wants into a place where they can recognize one another. His first achievement is to make himself useful. His difficulty, and sometimes his temptation, begins when a successful introduction would make him unnecessary.
The cost of finding includes more than the hours spent looking. A firm leaves work undone while a vacancy remains open; a worker spends savings while an employer searches elsewhere. A household visits houses that could never suit it because the descriptions conceal what matters. Some inquiries end in refusals, some promising leads arrive too late, and some possible trades never become visible enough to be refused. A competent matcher can reduce these losses by collecting information once that each party would otherwise have to collect separately. He does not have to create the house or improve the worker to create value. It is enough to make a possibility available that was previously beyond the parties’ practical reach. The comparison is with what finding would actually require without his service, including the matches that would fail to happen.
Search-and-matching theory makes that interval part of the economy rather than an embarrassment to be omitted from it. In Pissarides’s account, the worker is looking for a suitable job and the firm for a suitable worker; neither side is merely waiting for the other to accept a posted wage. The characteristics have to fit, and the process takes time. Unemployed workers can therefore coexist with vacancies without either number proving that the other is fictitious. But a matching friction is one part of an explanation, not a name for every cause of unemployment. Firms must also decide whether a job is worth creating, workers and firms must agree on its terms, and existing jobs may disappear. A recruiter who shortens the search acts within that larger process. He cannot supply a profitable vacancy merely by making its advertisement easier to find.
A larger register need not, by itself, make each search faster. Suppose an invented matching technology produces fifty matches from a hundred workers and a hundred vacancies during a period. If it has constant returns to scale, doubling both sides produces a hundred matches: twice the activity, but the same number of matches per worker and per vacancy. Improving the technology can raise those rates; changing the balance between workers and vacancies can improve one side’s prospects while making the other’s more difficult. This is why the matcher’s boast that he has the largest book is incomplete. The book may contain the widest range of possibilities, or merely the longest queue of people competing for them. Its value depends on who is in it, how well the descriptions distinguish them, and how effectively a promising contact becomes an acceptable match.
Nor is the best match always the first one that can be made. An introduction that saves a week but produces a relationship that fails after a month may be worse than a slower, more discriminating search. The matcher must decide what to learn before recommending, which defects to disclose, and when a further inquiry is worth its cost. These decisions give substance to expertise: a list of names is not yet a judgment about their fit. They also make the service hard to inspect. A client can count the interviews and see the eventual transaction; he cannot as readily count the better possibilities never brought to his attention. The fee buys work whose most important omissions may leave no record. To assess it, one must ask what the matcher is paid to accomplish, and what he knows that the parties cannot cheaply verify.
The success of an introduction changes the economics of the next encounter. Before the buyer and seller meet, each may need the register, the inquiry and the recommendation. Afterwards, they may possess one another’s names, know one another’s requirements and be able to arrange the next exchange directly. The information has escaped the service that assembled it. A matcher paid only when the exchange passes through him therefore has a reason to keep the introduction incomplete: enough communication for the trade, but not enough for the parties to dispense with the channel. That reason does not establish what any particular business does. It identifies a conflict built into one way of charging. The service is rewarded for joining the parties, while the continuation of its revenue can depend on keeping something between them.
An invented pair of transactions makes the temptation visible. Suppose an initial search costs the matcher forty pounds and he charges sixty for a successful introduction. Once introduced, the parties can repeat their trade directly without further matching expense. If he leaves them able to do so, his profit from the two occasions is twenty pounds. If he can require the second trade to pass through a further step that costs him five pounds and carries another sixty-pound fee, his profit rises to seventy-five. The extra fifty-five is his gain from retaining the gate. The extra sixty paid by the clients is not, however, sixty pounds of resources destroyed: fifty-five is transferred to him and five is spent on the step. If the second trade would still occur under either arrangement, the added resource cost in this example is five. If the fee prevents a worthwhile trade, there is a further loss. The distinction matters because a charge can redistribute the gains before it destroys them.
The obvious answer is to pay for the introduction and allow it to finish its work. Yet an upfront price has its own difficulty: the client must pay before knowing whether the matcher can find anyone suitable. A charge contingent on success shifts some of that risk to the matcher and makes his ability less expensive to test. It can also finance the searches that fail from the ones that succeed. A business that charged only the visible cost of its easiest completed introduction might never recover the expense of maintaining the register or investigating the unsuccessful cases. Even a later payment can be part of an agreed price for work already done. The relevant question is whether the arrangement finances useful finding on terms the parties could assess, or preserves an obstacle whose remaining purpose is to make the fee unavoidable. The date on the invoice cannot decide that by itself.
A commission can produce the opposite distortion: the matcher may want the search to end too soon. Suppose, in an invented example, that waiting an extra week would raise a sale price by ten thousand pounds with certainty. An agent receiving two per cent gains only two hundred from waiting. If that week costs him three hundred, he would prefer to close now. The owner, who retains ninety-eight per cent of the improvement, gains ninety-eight hundred before his own waiting costs; with five hundred of such costs, he would prefer to wait. Here both parties desire a sale, both would like a higher price, and the commission rises with it. Their disagreement survives all three facts because the costs and the marginal proceeds fall differently. Alignment in the direction of a reward does not establish alignment in the decision it is supposed to guide.
Better information gives the matcher a means of acting on that disagreement. He may know that a further offer is plausible when the owner can see only the one in hand. Advice to accept can then sound like an expert assessment of the market while also serving the adviser’s desire to move to another client. Dispensing with the expert is not necessarily the answer: the uninformed owner may do worse alone, even after allowing for the expert’s incentives. Levitt and Syverson make precisely this possibility part of their account. The difficult choice is between useful but interested advice and a poorer source of information, not between manipulation and effortless knowledge. A second opinion, a more informative record of offers, or a contract that rewards a different outcome has value only insofar as it improves that actual choice.
The matcher can also have good reasons to remain after the introduction. Requirements change, new candidates arrive, a promised quality must be checked, and a failed relationship may need replacing. A payment service or a guarantee may make continued use of the channel preferable even when the parties know how to contact one another. In that case recurrence purchases something recurring. The comparison must allow for those services instead of treating direct contact as a complete substitute for them. But bundling makes the judgment difficult: the same fee can pay for a new assurance and for permission to use an old acquaintance. The parties should be able to distinguish the two in substance, even where the invoice combines them. Otherwise the value of one useful function becomes the explanation offered for every restriction surrounding it.
Competition among matchers helps only if an alternative can perform the relevant introduction. A second office with an empty register does not give the same choice as a second office with access to suitable counterparties. Yet a large register does not by itself establish an unassailable position. Parties may consult several services, compare information, refer one another directly, or move together to a better venue. Each route has costs, and the pricing of entry and successful exchange can affect whether it is worth using. Caillaud and Jullien’s matching model makes this dependence unusually clear: the possibility of using more than one intermediary changes both the service available and the strategies by which firms compete. Merely counting the firms, or observing that joining a rival is permitted, leaves the economically important part of the question unanswered.
What would count as the matcher doing his job well is therefore larger than a rising number of paid introductions. It includes suitable matches that last, searches that end when further searching is wasteful, and useful knowledge that the client can retain. These outcomes can conflict with one another: a cautious inquiry may delay a good offer, and a freely reusable introduction may make some costly searches harder to finance. The task is to arrange payment so that these conflicts are confronted rather than concealed in the word success. A fee for a completed match, a charge for access, and a continuing share of transactions reward different conduct. None deserves its verdict from its name. The question is what happens to the parties’ opportunities when the matcher can earn by improving them, and when he can earn by narrowing them.
The finest introduction gives the parties a way to meet again; the finest gate gives them no way to meet elsewhere.
The matchmaker’s strongest claim is the trade that would not have happened without him. It remains a strong claim even when his conduct requires scrutiny. But finding someone does not settle who should control the relationship thereafter, nor does creating a surplus settle how much of it the finder may take. The original question returns at the point of success: once he can keep the parties from meeting without him, what is he paid to do? A good answer lets the service earn enough to be supplied while making the continued fee answer to a continuing purpose. Then the knowledge assembled in the middle can become a connection between the ends. The next plate takes up a different need: the parties may already know where to meet, but one wants to trade before the other arrives. The dealer sells the time between them.