The People Who Hold the Spare Key

The town can afford the pump.

This is welcome news. The old equipment has become unreliable, and a replacement would make an important part of the water system easier to run. There are quotations on the table. The specifications have been checked by people qualified to do that work. One offer has a lower purchase price.

Then the maintenance supervisor asks about the spare parts.

A part needed for a common repair must be ordered from overseas. Delivery can take time. Training is available but costs extra. The service agreement does not include every visit. A backup power arrangement needs attention. The town has enough money to buy the pump and less money than it needs to make the promise behind the pump dependable.

The purchasing problem is familiar. We often budget for an object and hope that the service will take care of itself.

The meeting could end with the lowest price accepted and a photograph of the new machine. The difficult questions would then arrive later, under worse conditions, when residents were already waiting. Or the people in the room could admit that the purchase is only part of the decision and work out what dependable water service requires.

That second meeting is less likely to produce a triumphant photograph. It is more likely to produce a useful pump.

Our household fan began with a delivery window. By now, we can see how many things that window left out. Forecasts, parts, cash, ports, routes, records and reserve stock all helped make it possible. The public pump raises the stakes but does not change the basic question: what must keep working for the thing we bought to continue doing the thing we need?

Pay for the service you mean

A low price can be a genuine achievement. It can also be an incomplete price.

Suppose two offers for the town’s equipment differ by $3,000. The cheaper one may be just as good. Perhaps the seller is more efficient or has a better product. It would be foolish to assume that a higher price is evidence of virtue.

Now suppose the more expensive offer includes training, a suitable set of spares and an enforceable maintenance service that the cheaper offer excludes. The quotations describe different purchases. Comparing only their totals rewards the offer that leaves more work outside the frame.

The buyer must decide which services are actually needed and compare offers on the same basis. It should also examine the claim that an included service will be available. Who performs the work? When? With which parts? What happens if the named technician is unavailable? A service described in reassuring language can be as empty as an unstocked warehouse.

Life-cycle costing tries to widen this view beyond the initial price to costs over the period of use, including operation, maintenance and the end of the asset’s life. OECD work on public procurement describes both the value of this approach and the practical difficulty of using it well.¹

The calculation needs assumptions. Equipment may last longer or fail earlier than expected. Energy prices can change. A maintenance provider can perform poorly. Including a number for each future year does not make the future certain. It makes the assumptions visible enough to examine.

The town can ask for ranges, identify the largest uncertainties and see whether the preferred offer changes under reasonable alternatives. It can distinguish a cost that is likely to recur from a rare but severe failure. It can ask which promise is secured by a contract and which remains an optimistic statement in a sales presentation.

The work is worth doing because the cheapest purchase can become an expensive service. A machine that cannot be repaired locally may need long downtime after a minor fault. An unfamiliar design may require special tools. A lower-quality part may need more frequent replacement. On the other hand, an expensive package may contain services the town can competently provide for less.

The goal is a fair comparison, not a ritual preference for spending more.

There is also a public cost when a service fails. Households spend time arranging alternatives. Businesses lose work. People with the least money may face the hardest choices. Some of these costs will never appear in the equipment department’s accounts. They remain part of the decision whether or not the spreadsheet has a convenient cell for them.

A serious purchasing process makes room for the people who use and maintain the service. They can identify failures that a distant buyer might overlook. Their involvement does not replace technical expertise or financial discipline. It improves the description of the job those disciplines are being asked to do.

A reserve is an option with a bill

The town considers holding a critical spare nearby. It also considers paying a service firm to keep a trained team available. Both could reduce downtime. Neither is free.

A reserve gives the owner an option: if a specified need arises, something useful is already available. The payment can take the form of stock, storage, training, scheduled practice or a fee for capacity. Sometimes the option is never exercised. That does not settle whether buying it was wise.

Suppose, for a deliberately simplified calculation, a business pays $2,000 a year to keep a qualified backup ready. If a particular interruption occurs, that backup would avoid a $20,000 loss. Ignoring all other effects and assuming the backup works as intended, the annual fee equals one tenth of that avoided loss. A rough break-even calculation would compare the fee with the chance of the interruption multiplied by its avoidable cost.

This is a way to organize a question, not a way to pretend that the chance is known. A company may have too little history to estimate a rare event well. Several losses may occur together. The backup may fail under the same conditions as the main source. Some consequences, especially injury or loss of an essential service, cannot be reduced adequately to the firm’s private financial loss.

The calculation can still reveal a bad argument. If somebody says a $2,000 readiness fee is too expensive, ask compared with what. If somebody says the fee must be worth paying because the possible loss is large, ask whether the option would actually prevent that loss and whether a better alternative exists.

For difficult-to-predict events, it can help to work backward from consequences. Which failures would be tolerable for a day, a week or a month? Which would cause serious harm almost at once? Which can be managed with stock, which require another source, and which need a change in the service itself? The time-to-survive and time-to-recover work discussed earlier gives this reasoning a practical form.

A full insurance policy and a physical spare also do different jobs. Money paid after a loss may help a business recover. It does not necessarily restore service while a part is still unavailable. A spare may restore service but leave other financial losses uncovered. Treating all protection as interchangeable can leave a gap precisely where the buyer expected coverage.

The town therefore needs a small set of concrete promises. The spare must be correct and kept usable. Someone must be able to reach it. The repair team must have the necessary competence and safe access. The agreement must say when the team is available and what happens when several clients need help together.

A provider who has sold the same supposedly exclusive emergency capacity to every client may be able to serve one well and disappoint the rest. The buyer needs to know whether capacity is reserved, shared or merely requested on a best-efforts basis. Those can all be legitimate services at different prices. They should not share the same reassuring name.

The worker is not the spare part

A supply chain can look resilient because people keep rescuing it.

A warehouse worker stays late. A mechanic answers the phone on a day off. A driver absorbs another delay. A supplier’s staff work an extra shift after a buyer changes its order. Their effort can be generous, skilled and necessary. It can also conceal a system that has not paid for the service it routinely demands.

There is a dangerous convenience in calling people flexible. The word can describe autonomy, useful skill and varied work. It can also describe uncertainty imposed on somebody who has little power to refuse it.

If a business expects workers to be available at short notice, that availability has a cost. Someone arranges childcare, misses other work or cannot make plans. If the business regularly needs overtime, the staffing plan should acknowledge it. If employees are expected to spot danger, they need a practical way to speak and stop work without retaliation.

The ILO added a safe and healthy working environment to its framework of fundamental principles and rights at work in 2022. That recognition is a standard of obligation and purpose; it is not evidence that workplaces everywhere have become safe.²

The economic record requires similar care. An ILO review of Asian garment production in 2022 reported increases in real wages for most workers alongside persistent poor conditions, informality and vulnerability. Higher earnings and continuing harm can exist in the same industry at the same time.³

This complicates the easy stories. It is inadequate to say that export work is automatically exploitation. It is equally inadequate to point to jobs or rising wages and declare that the conditions need no further attention. People can value a job and want it to become safer, better paid or more predictable. Their choices should not be forced into the argument that outsiders find most convenient.

Nor should we assume that better conditions always require lower productivity. An independent evaluation of the ILO–IFC Better Work programme, conducted by Tufts researchers, found improvements in working conditions and evidence of better business outcomes in participating garment factories. Its research used several methods, including a randomized evaluation of supervisory training.⁴

That evidence does not promise a profit increase from every workplace reform. It does reject the claim that decent treatment and competent production must always be enemies. A supervisor who communicates clearly, a worker who can report a problem, and a schedule that allows a task to be done properly can improve the work itself.

The moral case is wider than the business case. We should protect people from serious harm even where the financial return is uncertain. But it is useful to challenge the lazy assumption that the cheapest apparent labor arrangement is necessarily the most efficient one.

Turn again to the public pump. If only one employee knows how to maintain it, that person becomes a single point of dependence. The answer is to train others, document the work and provide sensible cover. The answer is not to require that employee to remain permanently available, healthy and close to a phone.

A resilient service makes it possible for its people to take a day off.

This is more than kindness at the edge of an operational plan. Illness, retirement and ordinary life are predictable features of having human beings do work. A system that treats them as extraordinary surprises has misunderstood its own materials.

The buyer’s terms reach the workshop

The town’s service contractor has employees too. It needs enough steady work to keep them trained and employed. The town wants low prices, quick response and freedom to cancel. The contractor wants a commitment that makes readiness possible.

Their interests need not be identical for an agreement to be fair. They do need to be stated.

Suppose the town asks the contractor to buy tools and hold specialized parts, then reserves the right to switch suppliers with almost no notice. The contractor may raise the price, decline the job, or accept the risk because it has few alternatives. If it accepts and later fails, the town may call the supplier unreliable. It should also examine the terms it offered.

A powerful buyer can create fragility through the way it purchases. Short notice, changing specifications, late payment and unrealistic deadlines can force a supplier to carry costs it cannot manage well. We should not attribute every supplier failure to the buyer; businesses also make their own mistakes. But a contract is part of the operating environment, not merely a legal wrapper around it.

The good bargain divides duties so that each side can reasonably perform them. If the buyer needs a reserve, it can pay for a defined reserve. If volumes are uncertain, the parties can decide who carries which part of the risk. If changes are likely, they can agree how changes will be priced and scheduled. The alternatives should be compared under the applicable law and with competent advice where needed.

This sounds less stirring than demanding that every supplier become resilient. It is also more honest. You cannot consistently purchase the cheapest uncommitted service and expect the supplier to behave as though you financed a dedicated backup system.

The same problem appears in consumer promises. A seller advertises delivery at any hour with no visible charge, while the costs emerge through worker schedules, crowded routes or supplier terms. The customer may never have asked for those conditions. They are built into the offer before the customer sees it.

This is why consumer choice alone cannot carry the whole responsibility for supply-chain reform. Buyers need accurate options, but firms and public institutions choose the conditions under which those options are produced. A person buying a fan should be able to care about workers without being required to investigate every workshop behind a switch.

Information can help. So can enforceable standards, worker representation, credible inspection and purchasing practices that do not make compliance commercially impossible. The combination matters. A supplier code that promises excellent treatment while the buyer’s terms reward the opposite deserves close scrutiny.

It is easier to add a paragraph about values to a contract than to change the payment date. The payment date may tell us more.

The things nobody owns alone

Some of the most important parts of a supply chain serve many businesses at once.

A road, port approach, electricity network or emergency communication system cannot always be made reliable by one buyer negotiating harder with one supplier. The benefit of an improvement may spread across firms and households. The cost may fall on a public body whose budget is already strained.

This is where resilience becomes a public decision about shared infrastructure.

Imagine that several firms could avoid repeated delays if a short access road were repaired. Each has an incentive to hope someone else will pay. If nobody acts, all continue to lose time. A coordinated investment may create benefits no single firm could capture enough of to justify doing it alone.

That is a reason for public action or cooperation. It is not a reason to approve every project carrying the word strategic.

The proposal still needs evidence of the problem, a plausible account of benefits, a comparison with alternatives and a maintenance plan. A new road can fail to improve service if the actual limit is elsewhere. A larger port can add capacity that local connections cannot use. A new warehouse can become a monument to inventory nobody has funded.

The people affected also need a place in the decision. More freight movement can bring noise, pollution, safety risks or displacement to nearby communities. Jobs and lower transport costs may be real benefits. The burdens should be counted and addressed rather than treated as an unfortunate footnote to everybody else’s efficiency.

A public project is strongest when it explains the service improvement in ordinary terms. How will it reduce missed deliveries, make recovery faster or preserve access during a disruption? Which users benefit? Who might bear a new burden? What will be monitored after the opening ceremony?

The answers do not need false precision. A range of plausible results is better than one confident number built from hidden assumptions. A modest repair with a clear benefit can be more valuable than an ambitious expansion whose main achievement is its announcement.

Maintenance deserves a fair contest with new construction. The old bridge that continues to work does not produce the same news as a new bridge. The benefit can be just as real. Preventing a failure leaves behind an absence, which is harder to photograph than a completed structure.

Public budgeting must learn to value that absence without making it an excuse for spending that cannot be checked. Records of condition, service interruptions, repair times and completed maintenance can make the work visible. They are less dramatic than disaster footage and more useful for avoiding the next disaster.

A fair price for ordinary continuity

Who should pay for the changes we have described?

There is no single answer. A business may recover a private benefit from fewer interruptions and pay from its own funds. Customers may choose a more dependable service at a higher price. Public funding may support a service with broad benefits or protect access for people who cannot bear the full cost. Suppliers and buyers may share the cost of training, stock or capacity through a contract.

The source of the money should follow a clear account of the benefit and the obligation. It should not be chosen simply by finding the person with the least power to object.

Suppose a retailer adds a small amount to prices to finance better stock protection. Some customers may be willing to pay. Others have little room in their budgets. The change can improve reliability while making the product less affordable for those people. That is a genuine distributional issue, not proof that preparation is wrong.

For an essential service, a public subsidy or a different charging structure may be needed to preserve access. For an ordinary discretionary product, customers may reasonably accept slower delivery in exchange for a lower price. The important step is to avoid treating every form of reliability as equally necessary for every user.

We may not need a decorative item tomorrow morning. We may need a water-system repair part quickly. Giving both the same urgency can waste capacity that would be better used where waiting causes greater harm.

A business can offer choices without pretending that the fastest is automatically best. A wider delivery window can make routes easier to plan. Repair can be offered alongside replacement. A product with common service parts can be explained in terms a buyer understands. These options need to be real and reasonably priced, not punitive versions designed to steer everyone toward the familiar default.

Public buyers can also distinguish normal service from emergency service. If every order is marked urgent, the label stops carrying useful information. If urgent access is sold without enough capacity behind it, the promise becomes a competition in which somebody must lose.

One benefit of a clear service standard is that it permits an honest limit. A supplier can say what it can deliver under ordinary conditions, how it responds when conditions change, and which disruptions exceed its reserve. A customer can decide whether that protection is enough. The admission of a limit may create more trust than a promise no one could reasonably keep.

The world contains uncertainty. Contracts can divide it and preparation can reduce it. Neither can remove it by using a confident font.

The power to repair

There is another kind of payment hidden in the choice of product: what the buyer will have to pay to keep it working.

A device can be cheap to buy and difficult to repair. A part may be unavailable separately. The design may make access difficult. Necessary information may be restricted. A local technician may have the skill but lack the approved component. Sometimes those limits reflect genuine safety or technical needs. Sometimes they deserve to be challenged.

A buyer concerned with resilience should ask about repair before the failure. How long are parts expected to be available? Which parts are replaceable? What information and tools are needed? Who can perform the work safely? What happens if the original seller disappears?

For the town, these questions can shape the equipment choice. A slightly less sophisticated pump supported by local competence may provide better continuity than a more advanced model that cannot be serviced when needed. In another case, the newer technology’s reliability and efficiency may justify the added support requirements. The choice should follow evidence about the actual equipment and service, not nostalgia for simplicity or enthusiasm for novelty.

Repair also keeps skill in circulation. A technician who solves ordinary problems gains knowledge that can matter during unusual ones. A workshop that can earn a living from regular work is more likely to remain available when an emergency arrives. Occasional training without a viable practice may not preserve competence for long.

This is an argument for considering repair as part of economic capacity. It supports local activity without requiring every raw material or component to be made nearby. Imported parts, local skill and reliable information can combine to create a more independent user.

The independence lies in being able to act when something fails.

There are limits. Some products cannot be repaired economically or safely in every setting. A worn item may need replacement. A local improvised part can create risks if it has not been properly designed and checked. Respect for repair should not become permission to ignore engineering.

But a society that makes repair unnecessarily difficult gives up an important option. It asks people to depend more heavily on the arrival of new goods when a smaller intervention might restore the service they already own.

The spare part is a modest form of freedom. Its value depends on a person who knows how to use it.

Inventory of a repair

Set the right part on the bench.
Leave room for another pair of hands.
Write down what the old machine taught you.
Put the tool where it can be found.
There is work enough for morning.
Let someone go home tonight.

A promise that can be checked

The town’s meeting has become longer. That is not always a sign of wisdom. Meetings can expand without becoming more useful, and the participants still owe residents a decision.

They now have a better question to decide. Instead of asking only which pump to buy, they are choosing an equipment-and-service arrangement within a budget. They know the critical parts, the training required and the likely maintenance. They have identified a backup for the person who normally handles repairs. They have asked what happens if a shipment is late.

They still cannot prevent every failure. They can make the response less dependent on luck.

To keep the plan honest, they need a few measures that follow the service. How often is it unavailable? How long do repairs take? Which delays result from missing parts? Are scheduled checks actually completed? Are the people responsible trained and safely equipped? When a reserve is used, how is it replenished?

These measures are imperfect. A period with no failures can reflect good preparation or good fortune. A rise in reported faults can reflect deterioration or more honest reporting. The figures need interpretation, just as orders and port times did earlier in the book.

What matters is that the organization can learn. A near miss should prompt a useful review rather than a search for the weakest person to blame. A supplier’s warning should arrive early enough to change the plan. A worker’s concern should be heard before it becomes an incident report. A budget should be revised when the evidence shows that the original assumptions no longer hold.

The record should also make responsibilities visible. A public promise belongs to somebody who can be asked about it. The official need not personally turn every wrench, but cannot outsource the duty to know whether the service is being maintained. A contract is a tool for carrying out responsibility, not a machine for making responsibility disappear.

This is where I return to a concern that runs through my earlier work. In Virtue of the Gods, I wrote, “There are questions about care and concern for your fellow man.”⁵ Here, those questions arrive in less elevated clothing. They concern a shift roster, a storage room, a payment term and whether a person has the authority to stop unsafe work.

Care becomes credible when it changes what is provided.

That is a demanding standard because good intentions are cheaper than dependable arrangements. A company can praise workers while denying them time to work safely. A government can celebrate resilience while postponing maintenance. A customer can want ethical goods while being offered no reliable way to judge the claim. Each problem requires action at the level where the power actually sits.

We should not ask the person at the very end of a long chain to solve all of it through virtue at the checkout. Nor should we excuse the people with more power by saying the system is too complex to understand. Complexity makes complete control impossible. It does not make specific duties vanish.

The buyer can pay on time. The supplier can report a delay honestly. The employer can staff a service properly. The regulator can enforce a justified standard. The public body can maintain shared infrastructure. People can cooperate on a reserve rather than each assume someone else has one. None of these actions solves the world. Together, they change the conditions under which ordinary life proceeds.

Leave a workable instruction

The town chooses an offer it can sustain. It buys the equipment, the necessary training and a defined reserve of parts. It sets aside money for maintenance rather than treating the remaining budget as proof that the job is finished. The agreement identifies who responds, what is shared with other clients and which events require a different plan.

The final document is not glorious. It contains dates, duties and a few conditions that nobody hopes to use. The maintenance team has helped check it. A person who will take over next year can read what was agreed without having attended the meeting.

That last detail matters. Supply chains outlast the people who arranged them. Staff change. Businesses merge. The person who remembered why a particular part was kept on the top shelf retires. A service built entirely on unwritten knowledge can become fragile without any machine moving an inch.

We need to preserve the knowledge that lets another person act. This does not mean writing a manual so large that nobody can use it. It means leaving a clear account of the important dependencies, the warning signs, the available options and the people who can authorize a response. It means checking that the plan works when the usual expert is away.

The world arrives through work that most of us do not see. Seeing more of it should make us less casual about the people who carry it, more precise about the promises we buy, and less impressed by savings that merely send the bill elsewhere.

There will still be storms, mistakes, shortages and bad decisions. There will also be skill, invention and people who help one another recover. A dependable economy makes room for that help before it is desperately needed. It pays for some things to remain ready. It allows some things to stop. It keeps more than one person able to continue the work.

The meeting ends. Before everyone leaves, the maintenance supervisor asks one final question: if the person who usually opens the storeroom is away, who else has the key?

This time, someone can answer.

Sources and notes