Firm Boundaries and Business Relationships

The cheaper quote

Harbor Service is considering hiring an outside provider to perform a defined maintenance task. The provider quotes $42 per job. Harbor estimates that doing the same work internally costs $48. The purchasing manager proposes outsourcing and reports a prospective $6 saving.

Before approving the change, Harbor’s operations manager asks who will check the work, arrange return visits, and resolve disagreements about what the provider was supposed to do. Suppose expected additional rework borne by Harbor costs $8 per job, while administering and coordinating the contract adds another $3. The apparent saving has disappeared.

This does not mean outsourcing is generally a mistake. An outside specialist might have better equipment, deeper experience, or enough customers to use capacity more efficiently. The question is whether Harbor has compared the complete alternatives rather than an outside invoice with an internal cost estimate covering different responsibilities.

Earlier chapters examined how to organize work inside a firm. This chapter asks which work should be inside in the first place, and how to manage relationships that remain outside. Ownership, employment, and purchasing contracts allocate decisions and rewards differently. None makes coordination free.

Compare the same service on both sides

For this illustration, hold the task, volume, planning period, and required service standard constant. Harbor’s $48 internal estimate includes relevant avoidable operating costs, the opportunity cost of capacity, expected rework, and supervision. It is not merely the technician’s wage, nor an average cost padded with overhead that would remain after outsourcing.

The outside provider’s $42 quote does not include Harbor’s additional $8 of expected rework cost or $3 of contracting, coordination, and monitoring cost. The rework estimate is net of any reimbursement from the provider. Do not count a repair cost again if it is already covered by the quoted payment.

Assume transition costs are absent or already included and that other material consequences are comparable. Common unavoidable overhead and unchanged customer revenue cancel. Under those assumptions, the comparison is:

On a small screen, scroll the table sideways to see all columns.

The internal categories are included in the $48 total, not assumed to be zero. The example supplies no separate breakdown of that total. Outside cost is $42 + $8 + $3 = $53, so internal provision saves $53 − $48 = $5 per job.

Now change one assumption. If a better outside process reduces expected rework to $2 without changing the quote or administration cost, outside cost becomes $42 + $2 + $3 = $47. Outsourcing is then $1 cheaper. The break-even rework allowance is $48 − $42 − $3 = $3. Harbor should investigate whether the original rework estimate is credible, because that estimate can reverse the decision.

The $8 expectation is not an eight-percent failure rate. It could result from different combinations of failure probability and cost when failure occurs. Nor does an expected-cost comparison settle differences in severe downside risk or service reliability that have been left outside the numbers. Those differences would require an additional comparison.

Why not contract for everything—or own everything?

Finding a supplier, explaining the task, checking performance, and adapting an agreement all use resources. Economists call costs of arranging and governing exchange transaction costs. They are not simply the price paid for the service. A low invoice can coexist with expensive coordination.

Coase’s 1937 analysis asks why some transactions are organized through firms when markets already coordinate activity. His comparison includes the cost of using market transactions and the cost of organizing internally. The existence of one cost does not establish that the alternative is cheaper.

For Harbor, employing a service team can permit a supervisor to rearrange eligible assignments without negotiating a separate purchase for every adjustment. That authority has limits, and it does not supply perfect information. A supervisor can misallocate time, employees can respond to defective targets, and expanding the team can require additional management. Chapters 5–10 describe precisely these internal problems.

The outside provider faces a different set of incentives. Keeping the return from a cost saving may encourage better methods, but also shortcuts if quality is difficult to check. Harbor’s employees may face weaker direct cost incentives, but can be reassigned within their agreed responsibilities when an unusual problem occurs. Which difference matters depends on the work and the surrounding rules.

The boundary of the firm is therefore not the boundary between contracts and no contracts. Employment itself rests on an agreement. The economic distinction concerns which decisions are settled in advance, which can be directed within an ongoing organization, and which require agreement between separately controlled businesses.

What would make the comparison change?

Harbor should not extrapolate the $5 internal advantage to every service or scale of operation. An outside specialist may spread equipment and training costs across many customers. Harbor may avoid keeping rarely used expertise on staff. Those are possible production advantages, not automatic benefits of signing a contract.

Capacity matters too. Suppose a change in workload means doing this task internally now displaces an additional $6 of contribution from another job, beyond the capacity cost already counted. Internal relevant cost becomes $48 + $6 = $54, compared with the original outside cost of $53. Outsourcing now saves $54 − $53 = $1. Add only the new opportunity cost; do not count the same displaced work twice.

Time can change the choice in the other direction. Outsourcing might permit Harbor eventually to avoid equipment replacement or a supervisory position even if those expenditures cannot be avoided this month. Conversely, changing providers might require a temporary overlap of teams or costly data conversion. A short-run order decision and a long-run operating arrangement need not have the same answer.

Keep the comparison about a particular activity. Harbor could retain diagnosis and customer contact while buying a standardized testing service. It need not either perform every task itself or hand the whole customer relationship to another firm. But dividing the work creates handoffs, so a narrower contract is not automatically an improvement either.

A contract cannot settle every future adjustment

For routine work, Harbor can specify the task, the completion date, acceptable evidence, and the payment. If alternative providers can readily perform it, switching is a useful response to poor service. An occasional purchase may need little continuing coordination.

Other work is harder to specify. Suppose a provider encounters an unfamiliar fault and must choose between further diagnosis and replacing a component. Harbor may want the provider to exercise judgment, but cannot cheaply list every future condition and response. The parties may also understand what happened while finding it difficult to demonstrate the relevant facts to an outside reviewer.

An incomplete contract leaves some future actions, conditions, or adjustments unresolved. Incompleteness need not mean the contract was carelessly written. Specifying and verifying every detail can cost more than leaving some matters for later judgment. More pages do not necessarily produce a more useful agreement.

The managerial question is what happens when the unresolved matter arises. Who supplies information? Who may approve a change in scope? Can work continue while payment is disputed? Who decides whether the result meets the agreed standard? These are decision-rights questions across a boundary, not only purchasing questions.

A fixed payment gives the provider a reason to contain costs but can make additional work unattractive. Paying for additional effort can support adaptation while making it harder to distinguish necessary effort from an expanded bill. Harbor cannot solve that tradeoff merely by instructing the provider to be cooperative. It needs a feasible way to recognize changes and assign their consequences.

When an investment ties the parties together

Consider a separate Cedar Components transaction. An outside supplier is deciding whether to build a fixture, a device that holds a component in place during production. A fixture tailored to Cedar’s design would improve production, but would be worth much less in work for other customers.

This is a relationship-specific investment: its productive value depends substantially on continuing a particular relationship. The important comparison is its value in that relationship versus its value in the best alternative use, not simply how expensive it is.

A general-purpose machine can have a large purchase price yet remain useful to many customers. A small fixture can be highly specific if changing customers destroys most of its usefulness. Specialized training or adapting information systems can create a similar dependence without an unusual physical machine.

Specificity and sunk cost are different concepts. A cost is sunk when it cannot be recovered by changing the current decision. An asset is relationship-specific when redeploying it sacrifices productive value. A past expenditure can be sunk even when what it purchased remains broadly useful. Specificity matters here because leaving the relationship becomes a relatively unattractive alternative.

Before selecting a supplier, Cedar may receive several competitive offers. After one supplier has built the fixture and learned Cedar’s requirements, switching can become costly for Cedar as well. Dependence can develop on both sides even if there was vigorous competition at the beginning.

Williamson (2008) connects specific investments and the need for adaptation to the choice of contractual safeguards and organizational form. Cedar’s problem is not merely to select the lowest initial bid. It is to sustain worthwhile cooperation after the parties’ alternatives have changed.

Later bargaining can discourage earlier investment

Follow the fixture decision through time. Before construction, the supplier compares the investment cost with what it expects to earn over the relationship. After construction, much of that cost cannot be recovered, and the fixture has little value elsewhere.

Suppose Cedar can then press for a lower payment on terms the original agreement left open. The supplier might accept because continuing still earns more than its now-poor outside alternative. That does not establish that the initial investment was attractive on the revised terms.

The hold-up problem arises when vulnerability after a specific investment exposes its return to later bargaining, discouraging investment beforehand. A supplier anticipating this outcome may demand protection, choose a less specialized method, invest less, or decline the business. The loss can be an improvement that never happens, not just a visible dispute after money is spent.

The direction can reverse. If Cedar has reorganized its production around one supplier, the supplier may seek a higher payment when replacement would interrupt Cedar’s operations. The useful question is whose outside option has deteriorated, who controls the necessary assets, and which important terms remain open.

Not every renegotiation is hold-up. A genuine change in material costs or product requirements may make adaptation beneficial to both sides. Nor does a firm need to display overt dishonesty for anticipated bargaining to weaken investment incentives. Even an orderly negotiation over the future gains can leave an investor expecting too little return to undertake a valuable improvement.

Chapter 2’s sunk-cost rule still applies after the fixture is built: decide whether to continue by comparing future alternatives. Before building it, however, the expected terms of that later decision matter. “Ignore sunk costs” does not mean “ignore how today’s investment will affect tomorrow’s bargaining position.”

Safeguards have benefits and costs

Cedar and the supplier could use a longer agreement specifying initial volume, payment, change procedures, and what happens if the relationship ends early. The point is to make the expected return to investment more credible. But a rigid promise to purchase an obsolete component can preserve an investment at the expense of a valuable adaptation. A useful safeguard needs a way to deal with changed conditions.

Cedar could instead contribute to the fixture’s cost. That reduces the supplier’s initial exposure, but financing alone does not determine who may use the fixture, approve modifications, or take it elsewhere. Sharing a bill without assigning those decisions leaves the central problem unresolved. Cedar also needs a reason for the supplier to build and maintain an effective fixture after receiving support.

A more general design might reduce dependence by making the fixture easier to adapt to another customer. That improves the supplier’s alternative, but may sacrifice some production efficiency. Specificity can create gains; eliminating it is not always the best way to protect those gains.

On a small screen, scroll the table sideways to see all columns.

Reputation can support these arrangements. A party expecting valuable future business may forgo an immediate bargaining advantage to preserve cooperation. But that restraint depends on future gains, what others can observe, and the credibility of their response. The repeated-interaction logic from Chapter 4 does not guarantee cooperation when a relationship is ending or behavior is difficult to assess.

The table is not a ranking. For a standardized, easily replaced input, elaborate safeguards may cost more than they save. For repeated transactions involving specific assets and frequent changes, the same safeguards may protect enough value to justify their expense.

What ownership changes—and what it does not

Vertical integration brings successive activities, such as component production and assembly, under common ownership. Acquiring the supplier could give Cedar authority to coordinate linked assets and settle some disputes internally. It would not make the underlying allocation decisions disappear.

Ownership matters partly because contracts do not specify every permitted use of an asset. Residual control rights are the authority over uses left unspecified, within existing agreements and other constraints. Grossman and Hart (1986) emphasize both the benefits and costs of reallocating those rights: strengthening one party’s control can weaken another party’s incentive to invest.

For Cedar, owning the fixture could provide more control over where it is used. That is not the same as owning the supplier’s entire business. And owning either does not mean owning the workers or automatically obtaining their knowledge and cooperation. If good performance depends on the supplier’s continuing ingenuity, reducing its discretion and return may sacrifice part of what Cedar hoped to secure.

After an acquisition, disputes over delivery dates can become disputes between division managers. Headquarters must acquire information, set priorities, and assess results. The transfer-pricing problems in Chapter 10 now become relevant. Changing a supplier into a division replaces one set of coordination and incentive problems with another.

Cedar should therefore ask which specific decision common ownership would improve and whether a narrower arrangement could do that job. Buying the fixture while contracting for its operation is one possibility. A long-term supply agreement is another. Their feasibility depends on access, expertise, maintenance, and adaptation, not merely on whose name appears on an asset record.

Franchising: local ownership with shared responsibilities

Imagine Harbor wants to expand to another region. It could operate a branch with an employed manager, or consider a business-format franchise: an independent operator provides service using Harbor’s name and operating system under agreed conditions. This is a hypothetical expansion choice, not a fact about Harbor’s existing operations.

Suppose the franchise operator receives local customer revenue, pays operating costs and agreed fees, and retains the remainder. Better scheduling or reduced waste can increase that operator’s own earnings. Local ownership can therefore provide a strong reason to attend to details that headquarters finds costly to monitor. The operator also bears business risk; stronger incentives are not a free benefit.

The shared name introduces an external effect. If the operator saves money by reducing service quality, some customers may blame Harbor generally rather than distinguish that outlet. The operator receives the immediate saving but may bear only part of the damage to the wider reputation. Local profit alone may not provide enough incentive to protect a common brand.

Agreed service standards, training, audits, and complaint procedures can address this problem, but they reduce discretion and consume resources. An inflexible standard could obstruct a sensible response to local conditions. Harbor must distinguish the practices needed to protect shared quality from choices better left to the operator.

Incentives run in both directions. The operator relies on Harbor to maintain the operating system, training, and brand. If Harbor collects fees while neglecting those services, local effort may be worth less. Lafontaine’s (1992) study examines franchising through risk sharing and incentive problems involving both sides. The framework cautions against treating only the outlet operator as needing motivation.

A company-operated branch avoids neither weak effort nor reputation damage. It gives Harbor a different mix of direct authority, financial exposure, and managerial incentives. The comparison is between feasible packages of control and rewards, not between an unconstrained entrepreneur and a perfectly obedient employee.

Choose the arrangement, then specify how it works

For Harbor’s original task, the stipulated numbers favor internal provision. That recommendation changes when outside rework improves enough or internal capacity becomes more costly. For Cedar’s fixture, there is no supplied numerical answer. The important work is identifying dependence, investment incentives, and the decisions a proposed safeguard would change.

A useful recommendation should name the activity and the alternative, state the important cost and information assumptions, and identify what would make the recommendation reverse. It should also explain who checks performance, handles adaptation, and decides whether to continue the relationship. Choosing “buy” or “make” does not complete that design.

Chapter 12 combines this boundary choice with the internal choices developed throughout the book. A firm can select a sensible boundary and still undermine it through poor measures, rewards, or decision rights. Organizational fit requires considering the package together.

Study map: compare ways of organizing the same work

• Compare equivalent service and relevant costs, including coordination and capacity on both sides.
• Ask what a contract can specify and what remains for later judgment or verification.
• Identify investments that lose value outside the relationship. Distinguish specificity from sunk expenditure.
• Follow later bargaining back to its effect on the initial investment decision.
• Compare safeguards and ownership by the rights and incentives they change, including the new costs they introduce.
• In a franchise or other continuing relationship, examine both parties’ incentives and effects on shared reputation.
• Evaluate technology by its relative effect on feasible arrangements, not by a presumption that firms must grow or shrink.

Practice and discussion

1. Keep Harbor’s $42 quote, $3 contract-administration cost, and $48 internal cost. Calculate the expected outside rework cost at which the firm is indifferent. Then reduce only rework cost to $2. Which arrangement is cheaper, and by how much? Why does the original $8 estimate not identify a failure probability?
2. Return to the original outside total. An additional $6 of contribution would now be displaced by performing the job inside Harbor, beyond the capacity cost already counted. Recalculate the comparison. Explain why adding an allocation of unchanged headquarters overhead would not have the same meaning.
3. Cedar’s supplier can build a highly productive specialized fixture or a less productive general-purpose one. Explain how expected bargaining after installation could affect that choice. Propose a safeguard, specify one decision right it must assign, and identify a cost or risk it introduces. Would buying only the fixture solve every problem?
4. Compare a hypothetical Harbor franchise with a company-operated branch. Identify one local decision for which residual earnings could improve incentives, one effect on shared reputation, and one valuable action Harbor itself must have a reason to take. Explain what headquarters might lose by controlling every local decision.
5. An AI tool could reduce the cost of checking supplier work or the cost of coordinating employees. Explain why the first change might favor outsourcing while the second might favor internal provision. What information would you need if both costs fell? Identify one new cost that a proposal based only on time saved could overlook.

Check your arithmetic

For question 1, the break-even rework cost is $48 − $42 − $3 = $3. With rework at $2, outside cost is $47 and outsourcing saves $1. The costs of failures and their probabilities are not separately specified.

For question 2, internal cost becomes $48 + $6 = $54, so outsourcing at $53 saves $1. The added $6 represents newly forgone contribution, not a redistribution of unchanged overhead.

Source note

Coase frames the comparison of market and internal coordination; Williamson connects dependence and adaptation to governance; Grossman and Hart supply the residual-control perspective; Lafontaine supplies the two-sided incentive framing for franchising. Selected full-text passages, an institutional abstract, and an indexed published abstract were checked, not every result in those works. Harbor, Cedar’s fixture, both tables, and the public exercises are independent teaching constructions. This chapter offers economic reasoning, not contract drafting, employment classification, or merger advice.