The plate before this followed the gap of search to the dealer who fills it — the man who stands always ready to trade, so that the hurried need not hunt for a counterparty but may simply come to him. This plate looks at how that dealer is paid, and finds, at the heart of his trade, the figure that gives this whole volume its name. For the dealer does not post one price for the thing he deals in; he posts two. There is the price at which he will buy it from you, called the bid, and the higher price at which he will sell it to you, called the ask or the offer, and between them lies a gap — the spread — which the dealer quietly keeps. He does not earn his living by guessing which way the price will move; he earns it on the round trip, buying at the bid from the seller and selling at the ask to the buyer, and pocketing the difference however the price behaves in between. The thing, in his market, has no single price at all: it has two, and what one loosely calls the price is only the midpoint of a gap that the dealer holds open and lives within. The question this plate asks is the one the masthead would predict, now sharpened to its smallest and clearest form: what is that gap the price of? Is it the honest wage of a service the dealer renders, or the toll of a position he holds — and how, looking only at the two numbers he quotes, is anyone to tell?
Consider the mechanics plainly, for they are the model of every spread that follows. A dealer in some security stands willing, at any moment, to buy a quantity of it at his bid and to sell a quantity at his ask. A seller who must raise cash sells to him at the bid; a buyer who wants the security buys from him at the ask; and if the two come in close enough succession, the dealer has bought low and sold high without ever taking a view on the security at all, his profit the spread he set between his two prices. He is not a speculator betting on direction — though he bears, as we shall see, some risk of the direction moving against the inventory he is left holding. He is a provider of two-sided prices, a standing offer to take either side of anyone's trade, and his compensation is the gap between the side he takes and the side he gives. The familiar idea of a single market price, the one number that clears supply against demand, dissolves at this finest scale into two: the most anyone can sell for at this instant, and the least anyone can buy for, with an unbridgeable gap between them that is not a friction to be wished away but the very thing the dealer is in business to provide and to charge for. The law of one price, which a later plate will invoke against arbitrageurs, fails first and most intimately here, at the dealer's own quote.
What, then, is the gap the price of? The founding answer was given by Harold Demsetz, who saw that the spread is the price of immediacy. A trader who wants to deal has two ways open to him. He may be patient — post his own order at the price he hopes for, and wait for someone on the other side to come and take it, however long that takes. Or he may be impatient — demand to trade at once, this instant, without waiting, by accepting the price the dealer already stands ready to give. The patient trader, willing to wait, can avoid the spread, and may even earn it, by being the one who posts and waits rather than the one who crosses to the dealer's quote. The impatient trader pays the spread for the privilege of not waiting: he buys at the ask rather than waiting for a seller to appear, sells at the bid rather than waiting for a buyer. The spread, on this account, is precisely the price difference between trading now and trading eventually — the sum the hurried pay for immediacy, for the assurance of a counterparty reliably present at this moment instead of the costly and uncertain search for one who might come later, or never. Seen so, the dealer's gap is no toll but the wage of a genuine service, and a service the customer is free to decline: wait, and it costs nothing; insist on now, and the spread is its price.
The spread is the price of now: what the hurried trader pays for a counterparty reliably there this instant, instead of the wait and the search for one who might never come.
But immediacy is only the first and broadest name for what the gap contains, and the dealer's spread, examined closely, is found to be paying for three distinct things at once. There is, first, the plain cost of doing the business — the labour, the equipment, the exchange fees, the capital tied up in standing ready — which the spread must cover as any price must cover its seller's costs. There is, second, the risk the dealer runs in holding what he buys: having taken a security onto his own books at the bid, he must hold it until a buyer appears, and in the interval its price may fall, so that the spread must compensate him for bearing an inventory he did not want and cannot instantly shed. And there is, third and most subtle, the danger that the person trading with him knows something he does not — that the eager seller is selling because he has learned the security is about to fall, the eager buyer buying because he has learned it is about to rise — so that the dealer systematically loses to the informed and must widen his spread to recoup those losses from everyone else. Order-processing cost, inventory cost, and the cost of the informed trader: these are the three faces of the gap, and the plates that follow this one are given to turning each in its turn. Here it is enough to have named them, and to see that the single number of the spread is a sum of very different things.
With those three components in view, the case for the spread as an earned thing is strong, and the volume should state it at full strength before pressing the other side. Every one of the three is a real cost or a real service, not a contrivance. The immediacy the dealer sells is genuinely valued — the proof is that traders pay for it willingly when they might instead wait for nothing — and the readiness that supplies it is genuinely costly to maintain. The inventory risk he bears is real risk, of exactly the kind the volume on the premium found deserving of a wage. The losses to informed traders are real losses, which someone standing ready to trade with all comers cannot avoid and must somehow recover. A spread that exactly covered these three — the cost of the readiness, the wage of the inventory risk, the toll taken by the informed — would be no toll of the dealer's own at all, but the honest, competitive price of a service that the patient need not buy and the impatient are glad to. To this extent, and it is a large extent, the dealer's gap is a wage and not a levy: he is paid for being reliably there, for holding what he is handed, and for losing, gracefully and continually, to those who know more than he does.
The patient trader can refuse the spread and wait; that the impatient pay it gladly is the proof that immediacy is a service, and the gap its honest wage.
And yet the spread, like every cut in this volume, is set by the one who keeps it and paid by one who often cannot see what a fair keeping would be — and there the toll creeps in. For the three honest components fix only a floor beneath the spread, the least it must be to cover the dealer's costs and risks; they do not fix its ceiling. What sets the ceiling is competition and visibility. Where many dealers quote and every quote is seen, a dealer who sets his spread above the honest floor is undercut by a rival who sets his lower, and the spread is driven down toward the true cost of the service. But where a dealer is the only one a customer can reach, or where the market is dark enough that no one can compare quotes, the spread can be widened past its honest floor to whatever the customer's ignorance and impatience will bear, and the excess is pure toll — a charge not for the service but for the dealer's position as the one the customer cannot get around. This is not a hypothetical. The dealer markets of the past were notorious for spreads kept artificially wide, most infamously when the market-makers of a great electronic exchange were found to have quietly avoided the finer price increments that would have narrowed their quotes, holding the gap open by a tacit common understanding until daylight and competition were forced upon them and the spreads abruptly collapsed. The gap, it turned out, had been part service and part toll all along — and no one looking at the quote could have said in what proportion.
The bid-ask spread is, in this, the whole problem of the volume rendered in miniature and made almost tractable. It is a single, visible number, and yet it is a sum of the honest and the extracted in a proportion the number itself will not disclose. Part of it is the genuine price of immediacy, which the patient can decline; part is the genuine cost of the dealer's readiness, his inventory risk, his losses to the informed; and part, where his position permits, is the bare toll of being unavoidable. The customer who pays the spread cannot read its composition from the quote, and the dealer who sets it has every reason to let the immediacy and the costs be seen and the toll go unmentioned. Competition and transparency are what press the toll out of the gap, narrowing it toward the honest floor; monopoly and opacity are what let the toll back in, widening it toward the ceiling of what the trade will bear. The spread is thus not fixed by nature at any level, but is the running outcome of a contest between the dealer's wish to widen and the market's power to narrow — which is only the masthead's contest, between the earned and the extracted, carried down to the smallest unit of trade. The plates that follow take up the gap's three honest faces in turn, to measure how much of any real spread each can account for; what they cannot account for is the dealer's margin of power, and that residue is the toll.
So the dealer's two prices, and the quiet gap between them, are the coin this volume has been turning since its first page — service on the one face, toll on the other, and the two so alloyed that the quote which shows the gap cannot show its composition. The plates ahead will assay the coin: the immediacy the dealer sells, and what that service of trading-now is truly worth; the inventory he is left holding, and what the bearing of it costs; the informed trader he cannot tell from the rest, and what must be charged against the knowledge he lacks. Each turns one honest face of the spread to the light, and what survives the turning — the part of the gap that no service, no cost, and no risk can explain — is the dealer's toll, levied for standing where the trade must pass. The next plate takes up the first and broadest of those faces, immediacy itself: the service of letting a man trade the instant he wishes, and the price the world puts on now. As the table says of every price, first to last: the argument remains open.