Chapter 1: The Exchange That Needs a Promise
An asset can have a price and still be difficult to sell in the amount, at the time, and on the terms its owner needs. The distinction becomes visible when many sellers want the same thing at once. A quote on a screen may remain available while the quantity behind it shrinks. A buyer may be willing to take a small position but unable to finance a much larger one. The market is still open; the arrangement connecting offers to completed trades is under strain.
In early 2020, investors seeking cash sold large volumes of U.S. Treasury securities and agency mortgage-backed securities. The Federal Reserve later reported wider bid-ask spreads and falling market depth. Its account identified several pressures, including the possibility that dealers faced constraints on the additional securities they could hold, and described policy actions taken to support market functioning.[O-1] This was a particular episode in particular financial markets, not a rule that every market will react in the same way. It does show why the ability to trade cannot be understood from a price alone.
A market is an arrangement for making possible, recording and enforcing exchanges among people who do not necessarily know one another. Prices matter because they can communicate changing conditions; they do not contain every relevant fact. A price may omit damage borne by a third party, reflect bargaining power as well as scarcity, or change before a worker or supplier has time to respond. Nor does every useful exchange require a central marketplace. The practical question is which participants can enter, what information and resources they control, how obligations are settled and who bears the cost when the arrangement fails.
This book follows those questions from ordinary transactions to supply chains, electronic platforms and financial markets. It treats a market as a structure people build and maintain, rather than a single organism with a will of its own. Competition can reveal alternatives and discipline a seller; it can also be limited by switching costs, ownership of an essential route or rules that exclude particular participants. Coordination by an institution may repair one problem while creating another. Each claim requires a mechanism, evidence and a countercase.
The aim is not to decide in advance whether markets are good or bad. It is to see what a particular market makes possible, for whom and under which conditions. When those conditions change, the same price can describe a very different set of choices.
Notes and Sources
[O-1] Board of Governors of the Federal Reserve System, Monetary Policy Report, June 12, 2020, “Federal Reserve Actions to Ensure Smooth Functioning of Treasury and MBS Markets,” https://www.federalreserve.gov/monetarypolicy/2020-06-mpr-part2.htm, accessed November 21, 2026. Selected explanatory passages were reviewed; figures, transaction-level data and the complete report were not. The report presents dealer balance-sheet constraints as one possible contributing factor.
The buyer has inspected the machine. The seller has agreed to a price. The delivery truck is booked for Monday, and the buyer has arranged a transfer from a bank account. The agreement looks complete until the parties ask what happens between those events. The seller wants payment before releasing the machine. The buyer wants the machine before sending payment. Neither request is irrational. Each reflects the possibility that the other party will not perform.
The following hypothetical transaction shows that timing can matter even without fraud. The seller may miss the truck because a worker is ill. The buyer's bank may delay a transfer for a routine check. The machine may arrive damaged, or the parties may disagree about whether a fault existed before the sale. A price describes what they have agreed to exchange. It does not by itself tell them who must act first, how performance will be checked or what happens if the sequence breaks.
They can solve the problem in several ways. The buyer can pay in advance and accept the risk of waiting. The seller can deliver on credit and accept the risk of nonpayment. A third party can hold the payment until a specified condition is met. The parties can make a smaller first transaction and use its result to decide whether to trade again. Each arrangement shifts risk, requires work and leaves something unresolved. An escrow service may reduce one party's exposure while adding a fee and a dispute over the condition for release. A small trial can reveal some information without showing how the seller will perform on a larger order.
The market is not the moment at which the buyer and seller discover that they both prefer a trade. It also includes the arrangements that make the trade possible. Transport, payment, records, warranties and a process for resolving disagreement determine whether the proposed price can become a completed transaction. They can be supplied by the parties themselves, by private intermediaries or by public institutions. Their relative importance changes with the product and the relationship. A cash purchase at a familiar shop requires less documentation than a shipment of specialized equipment between firms in different jurisdictions.
The cost of making a promise credible
A promise is credible when the person relying on it has some reason to expect performance and some means of responding if performance does not occur. The reason may be familiarity, a written agreement, a deposit, a reputation among other customers or the prospect of a future relationship. None works in every case. A seller can value a reputation and still lack the resources to fulfill an order after a supplier fails. A written contract can identify an obligation while leaving the buyer unable to afford the time and cost of enforcing it.
Imagine that the machine seller offers a warranty. The offer matters only if the buyer understands its duration, the faults it covers and the procedure for making a claim. A warranty from a firm that disappears after the sale is different from one supported by a viable service organization. A broad promise with a costly inspection requirement may be less useful than a narrower promise that can be exercised quickly. The buyer does not need to assume bad faith to ask who will perform the repair and how long the machine may be unavailable.
The seller faces a corresponding problem. A buyer may complain about a defect caused after delivery, or use the machine in a way the agreement did not anticipate. A reasonable process can protect both parties: record the machine's condition, define a period for inspection and specify who is authorized to approve a repair. These measures do not eliminate disagreement. They reduce the number of questions that must be reconstructed after the relationship has become contentious.
Documentation has a cost. A small purchase may not justify a formal inspection, and a complicated procedure can exclude buyers or sellers who lack time, access to a printer or familiarity with the language used. A large firm may be able to assign a specialist to the contract; a sole trader may be negotiating between other jobs. The same rule can protect both parties on paper while falling more heavily on one of them in practice. Good market design asks which risk a procedure reduces and whether a less burdensome way to reduce it is available.
Repetition changes the bargain
Repeated exchange can make a promise more credible because each party expects to meet the other again. A supplier may correct an error without disputing every detail because a continuing customer is valuable. A customer may accept a short delay from a supplier whose previous deliveries were reliable. But the expected relationship can also create dependence. A small supplier that relies on one large customer may be reluctant to challenge a late payment if losing the account would threaten its business. What appears to be trust from one side can feel like an inability to refuse from the other.
One recurring-order hypothetical concerns a baker who buys flour from the same distributor each week. The distributor knows the baker's order pattern and can arrange delivery before the shop opens. That knowledge reduces coordination work for both parties. When the distributor raises a fee, however, the baker cannot assess the change from the invoice alone. Another supplier might charge less but require a minimum order, deliver at an inconvenient time or require payment earlier. The useful comparison includes price, delivery, credit terms and the cost of switching.
A new supplier might improve the baker's position without replacing the old one. A small backup order can reveal whether the alternative actually delivers on time. Maintaining two relationships also has costs: lower volume may mean a higher unit price, and the baker must reconcile two invoices. Diversification is not free. Its value depends on the consequence of a missed delivery and the reliability of the alternative. If no other distributor serves the area, the baker may have to rely on negotiation, a collective purchasing arrangement or a different production schedule instead.
A single price cannot explain this relationship. It is the visible result of terms that include timing, credit, quality and access. Those terms are not fixed by nature; people and institutions write them, enforce them and sometimes change them. The next section will examine how a market records a claim when the goods, the payment and the evidence of ownership move on different schedules.
The interval between agreement and completion
A home sale offers a second hypothetical. The parties agree on a price, but ownership records, financing, inspections and the movement of money do not all change at once. The seller may need to repay an existing loan. The buyer may need a lender to release funds only after specified documents are signed. The records must identify the property and the people authorized to act. A mistake in any one step can delay the others. The sale is not simply a simultaneous exchange of a house for a stack of cash.
A closing process coordinates those dependencies. In U.S. consumer guidance reviewed in 2024, the Consumer Financial Protection Bureau describes different roles for an escrow officer, a closing or settlement agent, and, in some states, an attorney. The duties and participants vary by state.[1-1] That description does not establish the outcome of a particular transaction or explain every local rule. It shows that the work of exchanging funds and transferring ownership can be assigned to institutions other than the buyer and seller.
A buyer may see the closing fee as an expense added to the price. A seller may see the procedure as a delay before receiving money. Both observations can be true. The process also provides a way to coordinate actions that neither party wants to perform first without assurance about the other. The relevant comparison is not between a fee and a world with no cost. It is between the fee and the risks, delays and duplicated work involved in another feasible arrangement. A costly process can still be excessive or badly designed; calling it coordination does not make every charge justified.
The distinction between a transaction intermediary and a continuing escrow account matters. A person handling funds at closing may have a different role from a lender or servicer holding money for later taxes and insurance. Using the same word for both does not make their duties identical. A participant needs to know which money is being held, by whom, for what purpose and under what conditions it will be released. A general description cannot substitute for the agreement or the law governing the specific transaction.
Evidence that another person can use
A transaction produces information for the parties. The buyer learns whether the machine works; the seller learns whether the payment arrived. A market that supports repeated trade also needs some of that information to become usable by others. A receipt, title register, inspection report or review can make a claim easier to verify. Each is a record produced under rules about who can write it, who can see it and what an error means.
In a clearly labeled hypothetical, a buyer discovers that the machine's safety guard is missing after delivery. A dated photograph taken before shipment may help establish the condition in which it left the seller. A delivery record may establish who received it, but not whether the guard was present. A later inspection may identify the missing part without establishing when it was removed. These records answer different questions. A dispute becomes harder when a document is treated as evidence for something it was not designed to show.
Access to evidence is itself uneven. A large customer may insist on detailed records as a condition of purchase. A smaller customer may receive a standard invoice and have little time to examine it. A seller can also face a buyer whose internal approval process is opaque. The supplier may have performed the work but not know which document is needed to trigger payment. A record improves coordination only if the relevant person can obtain it, understand it and present it to someone with authority to respond.
A digital platform can make records easier to store and search, but it may also control access to them. A seller whose account is suspended may lose the practical ability to see transaction history, contact customers or challenge a decision through the same interface. An independent export, a reachable appeal process and clear retention rules can reduce that dependency. They also require maintenance and security. More information is not automatically better if it contains sensitive details that are widely accessible or retained indefinitely.
When the market is not the right substitute
Some exchanges work because the parties can specify what is being bought and when it has been delivered. Other arrangements are harder to reduce to a single purchase. A family caring for a relative may pay for transport or meals, yet the person receiving care may need continuity, choice and the ability to refuse a particular helper. A service contract can define hours and duties; it cannot by itself settle every question about consent or replace relationships that the participants value.
This does not mean that paid services are necessarily impersonal or that unpaid care is always fair. A skilled provider may offer reliable assistance and give a family member time to rest. An unpaid arrangement may be deeply valued, or it may depend on someone who cannot safely decline. The choice should be judged against actual alternatives and the preferences of the person receiving care. A price can make one cost visible without measuring every consequence.
There are also cases in which a transaction affects people who did not agree to it. A factory and a buyer can settle a purchase while a neighbor bears the cost of polluted water. The private agreement may be complete even though its effects are not confined to the contracting parties. Rules, liability, public monitoring or collective negotiation can bring some of those consequences into the decision. Each response needs its own design: a requirement that cannot be monitored may exist only on paper, while a poorly targeted restriction can impose costs without preventing the harm.
The contrast helps define the work a market does. It can allow people with different knowledge and preferences to coordinate without a single person directing every choice. It can also distribute bargaining power unevenly, miss effects outside the transaction and rely on institutions it does not itself create. The question is not whether a transaction has a price. It is whether the surrounding arrangements make that price meaningful and whether the people affected have a workable way to respond when something goes wrong.
Configure the escrow, not only the escrow amount
An escrow arrangement is credible only when its release conditions can be checked by someone other than the party that benefits from release. The agreement should identify the evidence, the deadline, the person who decides when evidence conflicts, and the route to appeal. It should also state what happens if neither party supplies enough information. A vague “satisfactory completion” condition postpones the hardest question until the relationship is already strained.
Escrow can reduce one risk while concentrating another. If one provider holds funds and records, its outage or insolvency may affect every transaction in the same system. A market designer can limit balances, segregate client funds where legally available, reconcile records, and arrange a backup process. Those safeguards have operating costs and do not remove legal or counterparty risk. A purchaser should understand who holds the money and what protection applies before treating a digital balance as equivalent to cash.
The agreement should distinguish a simple verification from a discretionary judgment. A delivery scan may prove that an item reached an address; it does not prove that the item met specifications. A laboratory report may show that a sample passed; it does not prove every unit was identical. If a neutral reviewer must make a judgment, the process should define scope, evidence, conflicts, time limits, and written reasons. The review should not be so expensive that it becomes unavailable for ordinary transactions.
A good settlement rule reduces the number of issues that must be decided after the fact. It cannot make the underlying facts certain. The institution should preserve enough evidence to reconstruct what happened and should return the parties to a workable position when its own process fails. In this way, settlement becomes a service with a measurable standard rather than a promise that funds will simply “be safe.”
Design the dispute path around error
The right to contest matters because institutions make mistakes. A dispute process should establish who can open a case, what records are visible, what evidence is required, who decides, how long it takes, and how to appeal. These are not afterthoughts. They affect whether people will participate and whether a dishonest party expects to get away with a breach.
A process that demands a long narrative, multiple documents, and a call during business hours may be accessible to a large firm but not an hourly worker or a low-income customer. Conversely, a process that grants every claim automatically can invite fraud and shift costs to other participants. Proportional procedures can distinguish a simple, reversible billing issue from a safety dispute or a pattern involving many transactions. They should preserve urgent service when delay would itself cause harm.
Complaint volume is an incomplete measure. A low rate can mean that service is good, that the problem is hard to recognize, or that filing is too costly. A better review combines complaint counts with resolution time, correction rate, repeat failures, abandonment, and outcomes by participant type. The organization should ask who never enters the process. If it cannot answer, its “low complaint rate” is not evidence that the exchange is fair.
A promise should be calibrated to capacity
A firm or public agency should not promise more than it can deliver under plausible stress. A broad uptime guarantee may depend on a single vendor. A warranty may depend on a supplier that can disappear. A city may promise continuity without having a backup operator. Honest qualification is better than a nominal guarantee that fails silently when needed.
This does not excuse weak performance. It clarifies the obligation. A promise can specify a service level, contingency, remedy, and party responsible for continuity. The institution should stress-test the chain, including a vendor failure, disputed record, and surge in demand. If a promise depends on another organization, the contract should create audit, notification, and replacement rights. Otherwise, the party making the promise may be unable to honor it while the participant has no practical alternative.
The value of an exchange includes the expected cost of nonperformance and its remedy. When a market lowers those costs, participants can transact at greater distance and scale. When it hides them, the apparent willingness to exchange may be supported by risks one side never chose.
The fallback is part of the bargain
A promise is tested most sharply when the ordinary path fails. If a payment processor is offline, can the buyer complete the purchase through another route? If a shipment is delayed, can the seller issue a refund without waiting for a carrier’s final report? If an identity service returns a false match, can the participant still access the service while correcting the record? These fallback arrangements determine whether a small operational failure becomes an exclusion event.
A continuity plan should identify the essential function, acceptable downtime, fallback authority, and record needed to reconcile later. A fallback can itself be abused: a manual override might bypass fraud controls, or an emergency exception might favor connected participants. Every override should have a reason, a time limit, and after-the-fact review. The point is not to eliminate discretion; it is to make the use of discretion traceable.
The order of remedies also matters. A system might first correct an inaccurate record, then restore service, then compensate for measurable loss. If it compensates without correcting the record, the person may face the same denial again. If it corrects the record but refuses interim access, the delay may make the eventual correction meaningless. The remedy should address both the source of error and the time-sensitive consequence.
A hypothetical online ticket exchange may suspend a seller account after an unusual login. The security control protects buyers, yet a false positive can strand valid tickets minutes before an event. A balanced design could offer a verified alternative login channel, preserve the ticket record, allow a rapid appeal, and protect buyers through a temporary hold on transfer. This adds operational expense. The firm must compare that expense with the costs of false denial and fraud, then publish the boundaries so that a user knows what to expect.
No institution can promise that every transaction will succeed. It can promise that a failure will be detected, assigned to an accountable party, and addressed through a known process. This is the difference between trusting a brand and relying on a system.
Notes and Sources
[1-1] Consumer Financial Protection Bureau, “Who should I expect to see at my mortgage closing?,” last reviewed February 2, 2024, https://www.consumerfinance.gov/ask-cfpb/who-should-i-expect-to-see-at-my-mortgage-closing-en-1903/, accessed November 21, 2026. Read the agency's short U.S. consumer guidance on closing participants and state-specific roles. It is not a review of individual closing documents or every jurisdiction's requirements.
