The dealer offers two prices where the market is usually said to have one. At the lower he will buy from you; at the higher he will sell to you. The distance between them is the bid-ask spread, and what it purchases is the right to transact before a natural counterparty happens to arrive. The seller need not wait for a buyer who wants exactly this quantity at exactly this moment. The dealer will take the other side, within the size and terms of his quote, and carry the problem of finding the next trader. He sells readiness. Whether the spread pays for that readiness or for the privilege of being the only person ready is the question the two prices leave open.
The service is easiest to see when the parties’ clocks differ. A holder needs cash this morning; someone else will have money to invest tomorrow. Their valuations might permit a trade, but their schedules do not yet permit them to make it. The dealer stands between the dates, buying now in the expectation of selling later. Demsetz made this waiting central to the cost of transacting. Orders flow into a market over time; a diagram in which supply and demand meet at a single instant conceals the expense of being available when they do not. The customer who wants immediate execution can pay someone else to wait. The customer who supplies an order and is willing to wait may compete to provide part of the same service.
A spread is a pair of offers, however, not a profit already earned. Consider an invented quote of ninety-nine to buy and one hundred and one to sell. A customer who buys a unit at the ask and immediately sells it at the unchanged bid loses two, before any separate fees. A dealer who buys a hundred units at ninety-nine and later sells them at one hundred and one earns two hundred before expenses. But if the eventual sale price is ninety-six, he loses three hundred. The displayed gap was two in each case when he bought. What he realizes depends on the prices at which the two trades actually occur, and on what happens between them. The quote promises a price to the arriving customer; it does not promise the dealer that the next customer will arrive before the world changes.
Inventory gives waiting a balance sheet. Money committed to a position cannot simultaneously support every other opportunity; a fall in value can exhaust resources before the position can be sold. A dealer who has bought more of an asset than he wishes to carry may make further purchases less attractive and encourage sales from his stock. He can also hedge, borrow, reduce the size he offers, or seek another dealer willing to take the exposure. Those responses change the cost and quality of the immediacy supplied. A small displayed spread supported for a negligible quantity is a different service from the same spread available for a large order. Even before information enters, there is no single cost of standing ready independent of the position, its financing and the amount the customer wants to trade.
Information makes the customer’s willingness to trade a possible warning about the price. The dealer may offer a perfectly sensible price on what he knows and still attract exactly the customers against whom that price is disadvantageous. A buyer who has better reason to expect a rise is more likely to accept the ask; a seller who has better reason to expect a fall is more likely to accept the bid. The dealer sees the order without seeing the knowledge behind it. A price based only on the asset’s unconditional expected value therefore neglects the information conveyed by someone’s willingness to take that particular side. The customer’s arrival is not merely an opportunity to transact. The direction of his trade can be evidence about the thing being traded, evidence the dealer must anticipate before he can observe it.
An invented one-order market strips the matter down. An asset will be worth either eighty or one hundred and twenty, with equal probabilities. One-fifth of arriving traders know which value will occur and buy in the high state or sell in the low one. The remaining four-fifths have no such information and are equally likely to buy or sell at the prices considered. The dealer cannot distinguish the two kinds. A buy order is then more likely in the high state: its arrival raises the probability of that state from one-half to three-fifths. Conditional expected value becomes one hundred and four. A sell order lowers it to two-fifths, making conditional expected value ninety-six. Those are the competitive ask and bid in this example. The spread of eight arises without inventory aversion, operating expense or a monopoly profit.
The gains and losses show whose problem the spread solves. The informed trader makes sixteen on his trade. The uninformed trader, whose direction reveals nothing about value, loses four in expectation. Weighted by their arrival shares, the dealer’s expected loss of one-fifth times sixteen is offset by his expected gain of four-fifths times four. He breaks even. Raise the informed share to two-fifths, leaving the other assumptions unchanged, and the ask becomes one hundred and eight while the bid falls to ninety-two: the spread doubles to sixteen, still without a dealer rent. A wider spread can therefore be the competitive consequence of a worse informational environment. It makes trading dearer for an uninformed customer even when it makes the dealer no richer in expectation.
Zero expected profit is a condition of this model, not a discovery about every dealer. Glosten and Milgrom isolate information by assuming a risk-neutral specialist with no transaction or inventory carrying costs and ample capacity to trade. Real capacity can bind, and the cost of financing it can change. Nor are processing, inventory and adverse selection three labeled coins that can simply be separated from any observed spread. A change in the mixture of customers can alter the position accumulated; a change in available capital can alter which orders are accepted; an improvement in information can alter trading itself. The categories explain possible sources of a price. Measuring their contributions requires a model and evidence adequate to the market in question. Calling the unexplained remainder rent would turn an incomplete account of costs into a finding about power.
The customer must also ask how much of a transaction the quote describes. Suppose an invented book offers a hundred units at one hundred and one and the next hundred at one hundred and two. An immediate purchase of a hundred and fifty units, with the offers unchanged and no other charges, costs fifteen thousand two hundred: an average of one hundred and one and one-third. The best ask of one hundred and one was true, but insufficient to price the whole order. Depth, the possibility of cancellation before execution, the terms of settlement and separate fees can all matter to the attainable result. Conversely, a negotiated or improved execution may occur inside a displayed spread. The two numbers at the top of the book are useful observations; they are not the entire price of every possible trade.
Waiting does not remove all these costs; it changes who bears them. A customer who posts an offer instead of demanding immediate execution may obtain a better price, or may never trade. While the offer remains available, new information can make it attractive to someone better informed. Meanwhile the customer may miss another use of the money or the asset. A comparison that treats the patient order as guaranteed execution at a better price has purchased its conclusion by erasing the service under dispute. To know what the dealer adds, compare the immediate trade with the uncertain sequence the customer would actually face without him. The relevant saving may be time, exposure, effort or the avoidance of a failed transaction. None is measured merely by wishing that buyer and seller had arrived together.
Power begins to enter when another provider could supply the same readiness on better terms but cannot reach the customer or obtain the necessary access. A dealer protected by a barrier can retain a gap that an equally informed, equally capable rival would narrow. Yet the number of dealers alone is a poor measure of that protection. A small market may support few costly inventories, while a market with many nominal dealers may direct orders in a way that gives a customer little effective choice. The inquiry is concrete: who can submit an executable competing offer, on what terms, for what size, and who is permitted to accept it? The privilege is valuable precisely because it interrupts the comparison that would otherwise test the price of the service.
Even an identified excess over the immediate expense of handling an order need not be an excess over the cost of supplying the service. Systems, staff and usable capital must be maintained between trades; some capacity earns its keep by being available in bad conditions when a simple average says it is idle. A return required to attract that capacity belongs in the comparison. At the same time, an appeal to readiness cannot validate any charge the market will bear. One must distinguish resources committed to serving orders from resources spent defending exclusion, and a normal prospective return from a surplus sustained by keeping alternatives out. The comparison is demanding because it must preserve the quality of the service while removing the privilege, rather than removing the service and calling the resulting absence cheap.
The dealer’s accounts and the customer’s welfare therefore answer different questions. A competitive dealer can transmit losses from informed trading to uninformed customers, as the invented example showed. Eliminating his profit would not eliminate that burden, because there was no profit to remove. A protected dealer can also provide a service worth more to the customer than the price charged, while charging more than an effective rival would require. Willingness to trade establishes neither that the environment is efficient nor that the division is fair. The spread must be read together with the information, capacity and alternatives that generate it. A verdict based on its width alone would condemn some costly services and excuse some cheap but protected ones.
The dealer’s two prices disclose the cost of crossing now; they do not disclose how much of that cost belongs to his position.
What widens the bid-ask past the cost of the service, then, is a failure of the alternatives that would contest it, after the risks and requirements of genuinely comparable service have been allowed for. Establishing that failure takes more than observing a large gap or a profitable day. It requires seeing which risks the dealer bears, which losses he passes on, and which rival offers the customer can actually use. The dealer is needed because the two sides do not arrive together. He earns the part of his return that makes their separation bearable. The claim becomes different when he can preserve the separation, or reserve the crossing, on terms no capable alternative is allowed to test. The next plate follows that position into the marketplace itself, where the crowd that makes exchange possible can become the gate that keeps it captive.