The driver enters the road having counted the minutes he expects to lose. He compares the journey with a train, a later departure, or staying where he is, and the delay ahead weighs in that choice. His arrival also lengthens the journeys of people already on the road. Their lost minutes belong to their accounts; he receives no bill for them. Congestion makes this division visible because everyone can suffer from a crowd to which each has chosen to add himself. The road may carry no entrance charge while exacting a substantial payment in time. William Vickrey's proposal was to bring the delay imposed on others into the price of the trip that causes it. A driver already bears his own congestion cost. The missing item is the addition his presence makes to the costs borne by the rest.
The marginal cost of road use includes both the new journey and the delay it adds to existing journeys. Imagine a fixed stretch of road over a specified period, with a steady flow represented by q. Let c(q) be the cost per trip, including operating expenses and travel time valued in money, the same for every traveler at that flow. Total travel cost is then q c(q). A small increase in flow adds a cost of c(q) + q c′(q): the first term belongs to the added travel, and the second records the higher cost spread over existing travel. An individual small enough to take the flow as given sees the first term in deciding whether to go. A toll of q c′(q) supplies the second. This is the distinction Vickrey drew between routes chosen according to the delay the traveler experiences and routes chosen while also paying for the delay imposed on others. If additional flow causes no additional delay, this congestion component is zero.1
A small numerical road shows how a privately worthwhile journey can cost more than it contributes. Measure flow in arbitrary units, and suppose the willingness to pay for the marginal trip is twenty minus q, while its private travel cost is two plus q. These are hypothetical schedules, with money-valued benefits that do not change with income, no other external effects, and no collection costs. Without a toll, travel expands to nine units: the marginal willingness to pay and the private cost both equal eleven. Yet the marginal social cost there is twenty, because the extra flow also raises the costs of the nine units already traveling. The marginal traveler receives a journey worth eleven while imposing a further nine on others. The same omission affects the journeys immediately below that margin, though by different amounts. The resulting traffic is an equilibrium of individual choices, with a volume above the point that maximizes the total benefit of travel less its total resource cost.2
A toll of six brings this road to its efficient flow of six. At that flow the private travel cost is eight and the marginal external cost is six; together they equal the marginal trip's value of fourteen. The toll confronts each traveler with that full cost, so that trips worth less than fourteen are forgone under the stated alternatives. The gain can be checked in the aggregate. Total benefit is twenty times the flow less half its square, and total resource cost is twice the flow plus its square. Their difference is fifty-four at six units of flow, against forty and a half at nine. Thirteen and a half is the improvement created by the changed travel decisions. The road remains busy, and each user still experiences congestion. Eliminating the last delay would require giving up journeys whose benefits exceed the costs they impose. The target is the point at which another journey ceases to add to the total surplus.3
The toll's receipts and the road's saved resources answer different questions. Six charged on each of six units produces thirty-six of revenue, but the payment transfers purchasing power to the recipient; the mere transfer adds nothing to the thirteen and a half of improvement. Indeed, travelers in this example have less consumer surplus after the charge. Their aggregate surplus falls from forty and a half to eighteen, while the recipient obtains thirty-six; adding the two yields the fifty-four already calculated. A retained traveler saves three in travel cost and pays six in toll, leaving a net private cost increase of three before any use of the receipts is considered. Thus a policy can improve this measure of total surplus while making its users worse off as a group before revenue is returned or spent. Distribution depends on who receives the money, which services it finances, and who can change routes, hours, or modes. The accounting makes those choices visible instead of settling them by calling the road efficient.4
The relevant charge follows the congestion a trip creates at its particular place and time. A road can be nearly empty at noon and crowded at eight; a delivery that obstructs a narrow street can delay others without traveling far. Vickrey's 1963 paper treated both time variation and the system of alternative routes as central to the pricing problem. A toll on one crossing may send traffic to another whose delay is still unpriced, so the effect on the whole network must be counted. The cost of collecting a charge also belongs in the comparison, especially if the method itself makes people stop. Capacity creates a further decision: a charge that guides use of the existing road does not by itself determine whether a wider road is worth building. Construction, land, maintenance, and the travel patterns that follow an expansion must be assessed separately. The simple road establishes the missing cost in one controlled setting; the city requires the same accounting across the choices that remain open.5
The highway puts the externality into a unit almost everyone recognizes: the minute spent waiting. Those minutes matter even when they appear in no cash ledger, and a toll can alter the decision that imposes them. Their conversion into a money-valued cost also exposes the limits of the calculation. Travelers differ in income, obligations, and available alternatives; a willingness to pay reflects these circumstances as well as the value of arriving. Judging access therefore requires a view of who should bear the cost and what choices people should have, alongside a calculation of how much delay a trip creates. The road can be held in common while its use is governed by an explicit rule. In The Boundary, that rule returns to the larger inquiry: how an institution brings a cost within the account of the person whose action imposes it.