> For the complete documentation index, see [llms.txt](https://laurence-wilse-samson.gitbook.io/textbooks/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://laurence-wilse-samson.gitbook.io/textbooks/financial-economics-claims-prices-holders/part-v-firms-as-issuers-of-claims/chapter_21_firm_contracts.md). # Chapter 21: The Firm, Contracts, and Institutions *Part V: Firms as Issuers of Claims — Financial Economics: Claims, Prices, and Holders* *** ## Opening Episode: Fisher Body, 1919 and 1926 In 1919 General Motors signed a ten-year contract with the Fisher Body Corporation of Detroit. GM agreed to buy substantially all of its closed automobile bodies from Fisher; Fisher agreed to supply them at a price equal to variable cost plus a markup of 17.6 percent, with protective clauses capping the price against what Fisher charged other automakers and against the average market price of comparable bodies. In the same year GM bought sixty percent of Fisher's stock. In 1926 it bought the rest, and Fisher Body became a division of General Motors, which it remained for the next fifty-eight years. The reason the contract was written at all is a fact about technology. An open touring car had a body that any competent metalworking shop could bend. A closed body — a steel roof over a wooden frame, with doors that had to seal — required large stamping presses and, more to the point, *dies*: hardened steel forms cut to the shape of one manufacturer's body panels. A die for a Chevrolet roof panel is worth a great deal to Chevrolet and close to nothing to anyone else. And the market for closed bodies was exploding. Closed bodies were roughly a tenth of GM's output in 1919 and something like two-thirds by the middle of the 1920s. Here is the story as economists have told it since Klein, Crawford and Alchian (1978). Before Fisher cut the dies, GM could shop among body makers, and competition among them held the contract price down to something near cost. After Fisher cut them, GM faced one supplier who could produce Chevrolet bodies and Fisher faced one buyer who wanted them. A market with many participants on each side had become, by the act of investment itself, a market with one on each side. In that situation the contract price is no longer disciplined by competition; it is whatever the two parties bargain to, and each has an incentive to threaten delay. The 1919 contract was an attempt to write the bargain down in advance. It did not work well. On the standard account, the cost-plus formula became far too generous once volumes rose, because the markup applied to variable cost and Fisher had little reason to economize on it; Fisher used relatively labor-intensive methods; and — the detail that made the story canonical — Fisher refused to locate body plants adjacent to GM's assembly plants, which would have saved transport and inventory but would also have created assets useless to Fisher if GM walked away. GM's answer in 1926 was to stop contracting and start owning. That is a very good story. It may not be true. Ronald Coase, who had first asked why firms exist at all, spent part of his ninetieth year arguing that the standard account of Fisher Body is largely invented. In a 2000 article in the *Journal of Law and Economics* he reported from the archives and from his own conversations with GM executives in the 1930s that GM never complained about plant location, that Fisher did in fact build body plants next to GM assembly operations, and that the 1926 acquisition was driven by GM's wish to secure the Fisher brothers' managerial services and to finance a very large expansion of body capacity. In the same issue, Casadesus-Masanell and Spulber argued that the 1919 arrangement was never an arm's-length contract between independent firms at all — GM already held control — and that the relationship shows cooperation rather than opportunism throughout. Freeland argued, from the same archives, that the integration *created* a hold-up problem rather than solving one, because after 1926 the Fisher brothers held control rights over GM that they used. Klein replied and did not concede. Two lessons, and this chapter needs both. The first is the concept. Whether or not Fisher held up General Motors, the mechanism the story illustrates is real, and it is the reason a large fraction of the world's economic activity takes place inside organizations rather than across markets. An investment that is worth much inside one relationship and little outside it converts competition into bilateral monopoly, and no contract written before the investment can specify what to do in every situation that might arise afterward. Everything in §§21.2 and 21.3 is an elaboration of that sentence. The second is a lesson about evidence, and it is one of the reasons this part of the book opens here rather than at a discounted cash flow. Corporate finance is unusually rich in canonical episodes that are repeated because they illustrate a theory well, and unusually poor in checks on whether the episodes happened. A reader who takes the Fisher story on faith learns the hold-up model. A reader who follows the 2000 symposium learns the hold-up model *and* learns what it costs to establish that a particular firm was held up by a particular supplier in a particular year — which is exactly the difficulty facing every empirical claim in Chapters 22 through 25. The question this chapter plants is Coase's, and it determines the subject matter of the rest of the book: why do some transactions happen inside firms and others across markets? The answer matters to a book about claims and holders for a reason that is not obvious until it is stated. **The boundary of the firm determines which cash flows are bundled together, and a security is a claim on a bundle.** When Fisher Body was independent, there were Fisher Body shares and General Motors shares, two separate streams of residual income with two separate sets of holders. After 1926 there was one. The firm is the technology that decides what there is to hold. *** ## 21.1 Coase: The Firm as the Supersession of the Price Mechanism Coase's 1937 paper opens by noticing something economics had managed not to notice. Economic theory of the period described a system coordinated by prices: a change in relative scarcity moves a price, the price moves resources, and no one directs anything. Yet look inside a factory and there are no prices. A worker moves from one task to another because a supervisor says so. Materials move from one department to the next by schedule, not by purchase. As Coase put it, borrowing D. H. Robertson's image, firms are "islands of conscious power in this ocean of unconscious co-operation, like lumps of butter coagulating in a pail of buttermilk." So the firm is defined by what it displaces. Inside a firm, the allocation of resources is achieved by direction rather than by contract at every margin — Coase's phrase is that the firm involves the **supersession of the price mechanism**. The employment relationship is the primitive case: an employee agrees, within limits, to do what she is told, in exchange for a wage, rather than negotiating a separate price for every task. Herbert Simon later formalized exactly this — an employment contract is the sale of an option over a set of future actions, which is a different object from a sales contract. Why would anyone want that? Coase's answer is that **using the price mechanism is costly**, and the costs are of three kinds: **Discovery.** Finding out what the relevant prices are. Someone has to learn which suppliers exist, what quality they deliver, and what they charge. This is not free, and the cost falls per transaction as the number of transactions rises, which is already a hint about firm size. **Negotiation and contracting.** Concluding a separate bargain for each exchange. Every contract consumes lawyers, attention, and time, and the cost is largely independent of the size of the transaction, which is why small transactions are the ones most likely to be internalized. **Enforcement and adaptation.** Making sure the other side performs, and adjusting when circumstances change. A long-term contract economizes on negotiation but must then say what happens under conditions no one anticipated when it was signed. Call the sum of these the **transaction costs** of using the market. If they were zero, there would be no reason for any transaction to occur inside a firm: every worker could be an independent contractor selling each hour to the highest bidder. That they are positive is the reason firms exist. But that argument, taken alone, proves too much. If organizing transactions internally avoids the costs of using the market, why is the whole economy not one firm? Coase's answer completes the theory by making it a margin rather than a direction. Organizing internally has costs of its own, and they rise with scale: an entrepreneur's attention is finite, the chance of misallocating a factor rises as more factors are under one direction, and the price of supplying factors may rise for a larger organization. So: > A firm will tend to expand until the cost of organizing an extra transaction within the firm becomes equal to the cost of carrying out the same transaction by means of an exchange on the open market, or to the cost of organizing it in another firm. That is the equilibrium condition, and it is the reason this literature is worth an economist's attention rather than a management theorist's. The firm's boundary is a margin at which two costs are equated, exactly as a firm's output is a margin at which two other costs are equated. Everything downstream — vertical integration, outsourcing, the make-or-buy decision, the choice between a bank loan and a bond — is a question about where a margin sits. What the 1937 paper delivered, then, is a question with the right shape and an answer that names its own missing piece. "Transaction costs" is a placeholder. It says the boundary sits where two unobserved costs are equal, without saying what makes one transaction costlier to organize by contract than another, and therefore without generating a prediction that can be taken to data. For thirty-five years the paper was cited more than it was used. The two literatures that follow are attempts to fill the placeholder in: Williamson by identifying the attribute of a transaction that drives its contracting cost, and Grossman, Hart and Moore by identifying what ownership actually confers. *** ## 21.2 Williamson: Asset Specificity and the Fundamental Transformation Oliver Williamson's answer is that the attribute which matters is **asset specificity**: the extent to which an investment supporting a transaction is worth more inside that relationship than in its next-best use. The Chevrolet die is the pure case. Specificity comes in recognizable forms — site specificity (a coal mine and a power station built next to each other, so that neither can move), physical asset specificity (the die), human asset specificity (a maintenance crew that knows one machine's faults), and dedicated assets (capacity built for one buyer's volume). The reason specificity is the right variable is the **fundamental transformation**. Before the specific investment is made, there may be many qualified suppliers, and the buyer can run an auction. Afterwards, the incumbent supplier holds an asset no rival has and no rival can cheaply build, so the buyer's alternatives are worse than they were and the supplier's alternatives are worse too. A large-numbers bidding situation at the outset becomes a small-numbers bargaining situation thereafter — and the transformation is produced by the parties' own efficient investment, not by any market imperfection present at the start. Bilateral monopoly would be a distributional matter if contracts could be written completely. They cannot, for reasons §21.3 makes central: the future is complicated, language is imprecise, and courts observe less than the parties do. So the terms of trade are open to renegotiation, and each side can threaten to delay in order to move them. The threat is the **hold-up problem**, and its cost is not the transfer itself. It is that a party who expects to be held up invests less than she should. Figure 21.1 is the whole argument in advance, drawn as the comparison §21.2 ends on: as specificity rises, the market's advantage in high-powered incentives is eaten by its disadvantage in adapting, and the transaction migrates first to a contract and then inside a firm. ![Figure 21.1: Make or buy](https://846781005-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F3EupdX99vVBoNySDtmxb%2Fuploads%2Fgit-blob-cacfe7ebd843dd7c10bd930111529b745f71551f%2Ffig_21_01_make_or_buy.png?alt=media) **Figure 21.1: Make or buy.** Governance cost against asset specificity, one curve per structure, with the lower envelope shaded by the structure that attains it. The theory asserts an ordering and not a set of levels: the market has the lowest bureaucratic cost at zero specificity and the hierarchy the highest, while the market's cost of maladaptation rises fastest in specificity and the hierarchy's slowest. Those two facts alone generate the two crossings, and the crossings partition the axis into Williamson's three regions — buy, contract, make. Nothing here is measured; k1 and k2 are solved from the drawn curves rather than placed by hand, and their positions carry no empirical content. What the figure claims is the shape: that the make-or-buy question has an interior answer on both sides, so that neither integration nor outsourcing is a direction of travel. *Source: Author's construction, after Williamson (1985).* ### A hold-up example Here is the smallest version, in the terms of the opening episode. All figures are in millions of dollars. A body maker can install dies specific to one automaker's design at a cost of 120. The dies have no value in any other use. Without them, the two firms can still trade — the body maker can supply open bodies with general-purpose tooling — and the joint surplus over the life of the relationship is 300. With them, the joint surplus is 500. Installing the dies therefore adds 200 of joint value at a cost of 120, and is efficient by a wide margin. Now suppose the price cannot be fixed in advance in a way that survives — the contract is silent about the situations that actually arise, or the formula in it turns out to be the wrong formula, which is what the standard account says happened to the 17.6 percent markup. Then, once the dies are sunk, price is set by bargaining, and the parties split the ex-post gains. Write $$\phi$$ for the share of the joint surplus the supplier captures in that bargain. The supplier installs the dies only if $$ \phi \times 200 \ge 120, \qquad \text{that is, } \phi \ge 0.6 $$ Table 21.1 works it through. **Table 21.1: Hold-up and the specific investment** | Supplier's bargaining share $$\phi$$ | Supplier's share of the 200 gain | Net of the 120 cost | Dies installed? | Joint surplus realized | | ------------------------------------ | -------------------------------- | ------------------- | --------------- | ---------------------- | | 0.25 | 50 | −70 | No | 300 | | 0.50 | 100 | −20 | No | 300 | | 0.60 | 120 | 0 | Indifferent | — | | 0.75 | 150 | +30 | Yes | 380 | | 1.00 | 200 | +80 | Yes | 380 | *Source: Author's calculation. Joint surplus is 300 without the dies and 500 with them; the investment costs 120 and is fully specific. "Joint surplus realized" is net of the investment cost.* Figure 21.2 puts the arithmetic on its timeline. Read the row that matters. Under an even split of the ex-post gains — the natural benchmark when two parties bargain face to face with nothing else to fall back on — the supplier captures 100 against a cost of 120 and does not invest. The relationship generates 300 instead of 380. Nobody behaved badly; nobody was fooled. The eighty of lost surplus is the price of an unwritten contract, and it is paid before any hold-up actually occurs. **The characteristic cost of hold-up is underinvestment, not expropriation.** Expropriation is a transfer; underinvestment is a deadweight loss, and it is what makes the boundary of the firm an efficiency question rather than a bargaining one. ![Figure 21.2: The hold-up problem](https://846781005-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F3EupdX99vVBoNySDtmxb%2Fuploads%2Fgit-blob-65e2d20cdd310bc49b372ed773b618f034a84fdc%2Ffig_21_02_hold_up.png?alt=media) **Figure 21.2: The hold-up problem.** Panel (a): the dates. The specific investment is sunk at date 1 and the price is agreed at date 3, and the whole of the problem lives in that order — before the die exists there are many suppliers and the buyer can run an auction; after it exists there is one, and Williamson's fundamental transformation has happened without anyone doing anything inefficient. Panel (b): joint surplus realised as a function of the supplier's ex-post bargaining share. The investment costs 120 and adds 200 of joint value, so it is made only when the supplier expects at least 120 of that gain — a threshold at phi\* = 0.60. Under the natural even split the supplier expects 100, the dies are not installed, and the relationship realises 300 rather than 380. The shaded block is the 80 that is never created: a deadweight loss rather than a transfer, and paid before any hold-up occurs. *Source: Author's calculation from the worked example of §21.2 and Table 21.1.* Notice also that the parties are not helpless. Anything that raises $$\phi$$ above 0.6 solves the problem — which is the point of the exercise. A long-term contract with a price formula does it if the formula holds up. A hostage does it: let the buyer make a specific investment of its own, and each side now has something to lose. Reciprocal exposure, take-or-pay clauses, posted collateral, reputation supported by repeated dealing, and arbitration clauses all work in the same direction. Vertical integration does it by removing the second party altogether. Williamson's contribution is to turn this into a comparative assessment rather than a story. **Governance structures** — the market, hybrid forms such as long-term contracts and franchising, and unified ownership — differ in their capacity to adapt to unforeseen change and in their bureaucratic cost, and transactions differ in specificity, uncertainty, and frequency. The prediction, which he called discriminating alignment, is that transactions are assigned to governance structures so as to economize on transaction costs: low specificity to the market, high specificity and high uncertainty to the firm, and the middle to hybrids. That is Figure 21.1's lower envelope, read as a rule for assigning transactions rather than as three curves. Integration is not free — an internal division cannot be fired, its managers lobby for capital, and the high-powered incentives of the market are blunted inside the hierarchy — which is why the answer is a margin and not a direction of travel. ### What the evidence says This is one of the more successfully tested propositions in organizational economics, and the tests are worth naming because they show what "measuring specificity" requires. Monteverde and Teece (1982) classified automotive components by the engineering effort specific to one manufacturer's design and found that components with more of it were more likely to be produced in-house. Joskow (1987) used coal contracts between mines and generating stations and found that contract duration rises with site specificity, mine-mouth plants signing the longest contracts. Baker and Hubbard (2004) used the diffusion of on-board computers in American trucking as a shock to what could be contracted upon, and found ownership patterns shifting exactly as the theory predicts when monitoring improves. Lafontaine and Slade's (2007) survey of the vertical-integration evidence reports that the results line up with the transaction-cost prediction with unusual consistency for an empirical literature in this area. The standing weakness is measurement: specificity is rarely observed directly, proxies are constructed by the researcher, and the direction of causation between organizational form and the investments made under it is exactly what the Fisher Body symposium fought about. *** ## 21.3 Incomplete Contracts and Residual Control Rights Williamson explains why parties want protection. He does not, on his own terms, explain what ownership *is* — why buying the supplier is a different thing from signing a very long contract with it. If the parties could write a contract specifying everything, integration would be redundant, because any allocation of authority a merger achieves could be written into a document. The answer comes from Sanford Grossman, Oliver Hart and John Moore, in two papers (Grossman and Hart 1986; Hart and Moore 1990), and it is the sharpest idea in this chapter. Start from the premise. A **complete contract** would specify an action for every state of the world. Real contracts are **incomplete**: some contingencies cannot be foreseen, some can be foreseen but not described precisely enough to be written down, and some can be described but not *verified* by a court, so that a clause conditioning on them cannot be enforced. Anything a court cannot verify is, for contracting purposes, unwritable — the parties may both observe that the buyer's design changed, but if a judge cannot, the contract cannot condition on it. Now the definition that does the work. Because contracts are incomplete, there will be decisions to make that no contract covers: whether to modify the tooling, whether to run a second shift, whether to sell the machine. **Residual control rights** are the rights to make exactly those decisions — the rights not specified by contract. And the GHM answer to the question "what is ownership?" is: > **Ownership of an asset is the possession of the residual control rights over it.** To own a machine is to be the person who decides what the machine does whenever no contract says. That is a definition of ownership entirely in terms of decision authority, and it separates ownership from the right to the income the asset produces, which is a separate thing and can be sold separately. Keep both halves in view — cash-flow rights and control rights — because §21.4 is about securities that split them. Residual control matters because it sets outside options, and outside options set the division of the ex-post surplus, and that division determines investment incentives. A party who owns the assets can walk away and still use them; a party who does not owns only her share of a bargain she must strike. So ownership raises $$\phi$$ for the owner and lowers it for the non-owner. Since investments here are non-contractible — effort, attention, the knowledge a manager builds up — the only instrument available for encouraging them is the assignment of ownership. The prescription follows: **give the assets to the party whose non-contractible investment contributes most to joint value.** ### The logic in numbers Two managers, F (bodies) and G (assembly), can each make a non-contractible investment of managerial effort costing 120 in the units of §21.2, and neither investment can be paid for on delivery because no court can verify it. F's effort raises the joint surplus of the relationship by 200; G's raises it by 180. The relationship generates 300 before either invests. Both investments are efficient in isolation. There is one critical asset — call it the tooling. Whoever owns it controls its use if the parties fail to agree, so the owner retains part of the value his own effort created even in disagreement while the non-owner retains none. Take the retained part to be half. Then an owner who adds 200 to joint value raises his own disagreement payoff by 100 and the gains from agreement by the remaining 100, of which he takes half in bargaining: 150 in all, or three-quarters. A non-owner adds 200 to the gains from agreement and nothing to his disagreement payoff, so he takes 100, or one-half. Those two fractions, three-quarters for an owner and one-half for a non-owner, are all the table uses. **Table 21.2: Ownership regimes in the Grossman-Hart-Moore example** | Regime | F's share of his 200 | F invests? | G's share of his 180 | G invests? | Total surplus net of costs | | --------------------------------------- | -------------------- | ---------- | -------------------- | ---------- | -------------------------- | | F owns the tooling | 150 | Yes | 90 | No | 380 | | G owns the tooling | 100 | No | 135 | Yes | 360 | | Joint ownership (both consent required) | 100 | No | 90 | No | 300 | | First best (both invest) | — | Yes | — | Yes | 440 | *Source: Author's calculation. Base surplus 300; F's investment adds 200 and G's adds 180, each at a cost of 120. An owner captures three-quarters of the value his own investment creates, a non-owner one-half.* Four readings, and each is a result in the literature. **Ownership matters, and it matters through investment.** The three feasible regimes deliver 380, 360, and 300. Nothing about the technology changed across the rows; only the identity of the residual claimant on the tooling did. **The right owner is the one whose investment matters most.** F ownership beats G ownership here because 200 exceeds 180. Had G's contribution been the larger, the ranking would reverse. This is the operational content of the theory, and it is why GHM predicts that firms integrate *toward* the indispensable party rather than in any general direction. **Ownership is a scarce instrument, so integration is never a free improvement.** Under F ownership, G's incentive is worse than it would be under G ownership. Every allocation of control that strengthens one party weakens the other, which is why 440 is not attainable and why the theory has a genuine boundary rather than a corner solution. This is the precise sense in which the transaction-cost tradition's "costs of internal organization" appear here as an equilibrium object. **Joint ownership is dominated.** Give each party a veto and neither retains anything on disagreement, so both are non-owners for incentive purposes and both under-invest. The prediction that joint control is generally worse than one-sided control is one of the more distinctive implications of the framework, and one that anyone designing a joint venture should have in front of them. *** ## 21.4 The Bridge: A Security Is an Allocation of Cash-Flow and Control Rights Now the payoff, and it is the reason this chapter opens a corporate finance part rather than an organizational economics course. Hold the GHM definition in mind: ownership is residual control, and residual control is separable from the right to income. Then look at what a corporation's balance sheet actually contains. Common equity carries a claim on whatever is left after everyone else has been paid — a residual cash-flow right — and the right to elect the board, which is residual *control*. A bank loan carries a fixed cash-flow right and a set of covenants, and what a covenant does is transfer decision rights to the lender in specified states: breach the leverage test and the lender can accelerate, block the dividend, force a sale, or extract a waiver at a price. A secured bond carries a fixed cash-flow right and a control right over one particular asset, exercisable in default. Preferred stock sits between, and a venture capitalist's convertible preferred contract splits cash-flow rights, board seats, liquidation rights, and vetoes along separate dimensions that the parties negotiate one by one. The generalization is the sentence this part of the book is built on: > **A security is an allocation of cash-flow rights and control rights across states of the world.** Not "a claim to a stream of payments." A claim to a stream of payments *and* an assignment of who decides, state by state. Chapter 3 priced payoff vectors and never asked who was steering the firm that generated them; that abstraction is the right one for pricing and the wrong one for design. Aghion and Bolton (1992) made the point formally: the optimal financial contract in an incomplete-contracts world is state-contingent control, with the entrepreneur in charge when things go well and the investor in charge when they do not, precisely because their objectives coincide in the first case and diverge in the second. Hart and Moore (1998) show why a *debt* contract in particular emerges from this logic — a fixed promise whose breach triggers a transfer of control is a way of making the entrepreneur's authority contingent on performance without needing a court to verify performance directly. That is exactly the pattern Table 21.3 records. **Table 21.3: Claims as bundles of cash-flow and control rights** | Claim | Cash-flow right | Control while solvent and in compliance | Control after default or breach | Developed in | | --------------------------------------- | -------------------------------- | ---------------------------------------------- | --------------------------------------------------------------- | -------------- | | Common equity | Residual | Elects the board; residual decision authority | Subordinated; often extinguished | Chs 12, 23, 24 | | Bank loan with covenants | Fixed, senior | None directly, but covenants constrain | Acceleration, waiver bargaining, veto over major actions | Chs 24, 25 | | Secured bond | Fixed, senior, collateralized | None | Right to foreclose on the specific collateral | Chs 10, 24 | | Unsecured bond | Fixed | None; indenture terms are looser than a loan's | Claim in bankruptcy; control through the reorganization process | Chs 10, 24 | | Preferred stock | Fixed dividend, senior to common | Usually none | Contingent voting rights on arrears; liquidation preference | Ch 23 | | Convertible preferred (venture capital) | Negotiated, convertible | Board seats, protective provisions, vetoes | Liquidation preference; control shifts on milestone failure | Chs 18, 25 | *Source: Author's summary of standard contractual terms. The empirical anatomy of the last row is Kaplan and Strömberg (2003), who show that venture capital contracts allocate cash-flow rights, board rights, voting rights, and liquidation rights separately and make each contingent on observable performance.* Read the table's third and fourth columns together and the theory of the firm has become the theory of capital structure. Leverage is not only a tax shield and a bankruptcy cost, which is how Chapter 23 §23.2 will first present it; it is a rule for reassigning residual control as a function of the state. Default is the state in which the assignment flips, and Chapter 10's default machinery — the barrier, the recovery rate, the distance to default — is the pricing counterpart of the same event. Covenants are the fine structure of the flip, and Chapter 24 §24.1 works through them. Governance is what the holders of residual control do with it once they have it. And not all of the allocation is negotiated: a large part of it is supplied by corporate law as **default rules** — the board's authority, the fiduciary standard, the voting and quorum rules that apply unless the charter says otherwise — which the parties may contract around at a cost and mostly do not. Residual control is therefore a legal object as well as an economic one, which is why Chapter 24 §24.3 can treat boards and voting as instruments without asking each firm to have invented them. One consequence deserves stating in advance, because it is easy to miss and it is the joint between Part V and Part IV. If a security is a bundle of cash-flow and control rights, then holders differ not only in what payoffs they want but in what rights they can use. An index fund holds residual control rights over thousands of firms and has no capacity or mandate to exercise most of them — which is Chapter 17's common-ownership and stewardship problem. A distressed-debt fund buys a bond largely *for* the control right that arrives in default. An insurer holds an investment-grade bond precisely because its capital charge assumes it will never have to exercise anything. Chapter 20 showed that holders' demand curves are shaped by their constraints; Table 21.3 says that the object those demand curves are written over has a control dimension as well as a payoff dimension, and that issuers choose both when they design a security. *** ## 21.5 Property Rights, Teams, and the Nexus of Contracts Three further strands complete the frame, compactly. **Team production and monitoring.** Alchian and Demsetz (1972) start from a different primitive: production in teams, where output is joint and individual contributions cannot be separated out of it. If output cannot be attributed, effort cannot be paid for, and everyone shirks. The solution is to hire a specialist monitor — and to solve the problem of who monitors the monitor by making that person the residual claimant on the team's output, with the rights to alter the team's membership and to sell that position. The firm, on this view, is not authority at all but a contractual arrangement centered on a residual claimant with metering rights. That the residual claimant is the party with the strongest incentive to monitor is a proposition Chapters 23 and 24 will lean on when they ask what debt does to managerial discipline. **The nexus of contracts.** Jensen and Meckling (1976) pushed the deflation further. The firm is a legal fiction serving as a nexus for a set of contracting relationships among individuals — employees, suppliers, customers, creditors, shareholders. It is not a person with objectives; asking what "the firm" maximizes is a category error, and what one can ask instead is what equilibrium emerges from the contracts. Their apparatus is the theory of **agency costs**: the monitoring expenditures of the principal, the bonding expenditures of the agent, and the residual loss that survives both. That apparatus is not developed here because Part V deploys it twice, where its prerequisites live — as the agency cost of debt in Chapter 23 §23.2 and of free cash flow in Chapter 23 §23.5, and as the perquisite model and the machinery of boards, compensation, blockholders and activism in Chapter 24 §§24.2-24.3. What matters at this point is only the relation between the two views. GHM's reply to the nexus-of-contracts position is that if the firm were nothing but a bundle of contracts, ownership would be empty; it is the incompleteness of the contracts, and therefore the existence of decisions no contract covers, that makes the firm a distinct object rather than a legal convenience. **Institutions.** Douglass North's larger point is that the same reasoning applies one level up: the enforceability of any contract depends on institutions — property rights, courts, the rule of law — and these are themselves the outcome of a political and historical process, which is why economies with similar technology and different institutions perform very differently. Chapter 25's cross-country variation in firm financing is where this book cashes that in. The same reasoning has also been carried well outside the make-or-buy problem it is usually taught with: Roumasset (1978) reads agricultural tenancy — sharecropping against fixed rent against wage labor — as the contractual form that economizes on the joint cost of enforcement, monitoring, and risk-bearing, exactly the calculation §21.2 runs over asset specificity. A framework that explains a tenancy contract in a rice economy and a component supply contract in Detroit with one apparatus is either general or empty, and the empirical record in both settings is what decides which. The intellectual history of the new institutional economics — its relation to the older institutionalist tradition, Coase's route from the 1937 paper to the 1960 social-cost paper, and the reception of all of it — belongs to the *History of Economic Thought* volume in this series and is not retold here. > **Box 21.1 — What incorporation actually does** > > The theory of the firm is usually taught with no legal content at all, which leaves a gap: the state does not supply "firms," it supplies a menu of legal forms, and the choice among them is part of the design problem. > > What a corporation supplies that a web of contracts cannot easily replicate is **asset partitioning**, in the sense of Hansmann and Kraakman (2000). Two directions matter. *Owner shielding* — limited liability — protects the shareholder's personal assets from the firm's creditors, and it can in principle be achieved by contract, one creditor at a time. *Entity shielding* runs the other way: it protects the firm's assets from the shareholders' personal creditors, and from a shareholder's demand to withdraw her share. That direction is the one contracts cannot cheaply replicate, because it requires binding creditors who are not parties to the deal, and it is what organizational law actually adds. > > The consequence for this book is direct. Entity shielding is what makes a firm's assets a stable pool that can be pledged, and a stable pool is what a security is a claim on. Limited liability is what makes an equity share a standardized, freely transferable instrument whose downside is bounded at zero — the property Chapter 8 relies on when it prices equity as a call option on firm value, and the property that makes anonymous public trading possible at all. *** ## 21.6 Why the Theory of the Firm Opens Corporate Finance The standard corporate finance sequence begins with the net present value rule. It is a good rule. But it takes as given the thing most in need of explanation: that there exists an entity which owns assets jointly, retains earnings, has a cost of capital, and issues securities against a bundle of cash flows. Every capital budgeting exercise presupposes a boundary that determines which cash flows are in the bundle, and the standard sequence never asks where the boundary came from. This book asks first, for three reasons. **The firm is a claim-issuing technology.** Part IV assembled the holders and showed that their constraints move prices. Part V asks where the claims they hold come from, and the answer is that firms manufacture them. But a firm can only issue claims against cash flows it has assembled inside its boundary, and the boundary is set by the transaction-cost and control-rights logic of this chapter. A vertically integrated automaker issues one equity claim on a combined cash flow; a disintegrated one issues two, with different risk, different covenants, different holders. When a firm spins off a division, sells a subsidiary, or securitizes a receivables book into a bankruptcy-remote vehicle, it is redrawing a boundary in order to change the menu of claims available to holders — and Chapter 3's 3Com and Palm episode is exactly that operation performed badly. The set of securities in existence is not exogenous. It is the residue of a long sequence of boundary decisions. **Capital structure is control-rights allocation.** Chapter 23 will present the capital structure question in the Modigliani-Miller frame: given the firm's assets, does the division of its cash flows into debt and equity affect its value, and if so through what friction? That framing is correct and incomplete, because the division of cash flows is simultaneously a division of control. Once §21.4's table is on the page, the pecking order reads as a statement about which claim's control implications are least costly to concede; a covenant package reads as a schedule of state-contingent authority; and the persistence of leverage ratios that Chapter 23 reports as an unexplained empirical regularity looks less strange, since a control allocation is a harder thing to revise than a financing mix. Chapter 24 then takes the frame as its subject outright. **Governance is what residual control does when no contract binds.** Boards, votes, blockholders, activists, and takeovers are only interesting because someone must decide the things the contracts left open. Chapter 22 §22.6's treatment of the takeover market as a governance mechanism, and Chapter 24 §24.3's treatment of boards and compensation, are both applications of a single proposition established here: control is allocated, allocation is costly to change, and the party holding residual control acts on its own objectives, which need not be joint value. There is a fourth reason, and it is the one that closes the circuit between the two halves of this book. Chapter 20 ended by answering Chapter 3's question — whose marginal utility prices claims? — with an ecology of constrained holders whose weights are measurable. That answer has a mirror image on the issuing side. Holders' demand is a demand for *particular securities*, distinguished by rating, duration, index membership, liquidity, and now by control content. Firms observe the prices those demands generate and respond by manufacturing the claims that are dear and retiring the claims that are cheap: Chapter 23 §23.7 develops that as gap-filling and safe-asset supply, and Chapter 25 §25.6 shows the same response in the choice between a bank loan, a syndicated credit, a bond, and private credit. Issuance is a supply curve meeting Part IV's demand curves, and the firm is the entity that can move along it — but only within the boundary this chapter explains. So Part V proceeds as follows. Chapter 22 asks what a firm's assets are worth and what it should invest in, since a claim-issuing technology must first have something to issue claims against. Chapter 23 asks how the resulting cash flows are divided into securities and paid out. Chapter 24 asks who decides, under what contractual constraints. Chapter 25 asks what firms actually do, which is not always what the first three chapters predict. Each of them is, in the terms established here, a question about the allocation of cash-flow and control rights — and each of them ends where Part IV began, at a holder deciding whether to hold the claim. *** ## 21.7 Who Holds the Security, and What Their Constraints Do to Its Price Every chapter in Parts II, III and V ends by asking who holds the claim it has described and what their constraints do to its price. This chapter can ask it one chapter early, because the answer is what the four chapters after it assume. Section 21.4 said that a security is an allocation of cash-flow rights and control rights across states of the world. An allocation exists not because someone drew it but because some balance sheet can hold both halves of it. Take the control half first. Exercising a residual control right costs something: a covenant is worth having only to a holder who will read the compliance certificate, price the waiver, and staff the workout. That is Alchian and Demsetz's monitor stated as a job description, and the balance sheets equipped for the job are few. A bank with a credit officer and a restructuring desk can hold a maintenance covenant; so can a private credit fund whose capital cannot be redeemed, and a venture partnership that will keep a board seat for seven years — which is why Table 21.3's finest-grained allocations sit in venture documents, not bond indentures. Monitoring is close to a fixed cost per borrower, so control rights collect on the few balance sheets that can spread it across a large enough position. Now the holders who cannot monitor. A collateralized loan obligation passing its coverage tests, a bond fund tracking an index, and an insurer holding paper against a capital charge are for this purpose one holder: none can sit in a workout, and none wants a contract that might require it to. The claims written for those balance sheets have the control content stripped out — incurrence covenants rather than maintenance covenants, an indenture rather than a credit agreement, a trustee rather than an agent bank. Chapter 24 §24.1 documents that migration in the leveraged loan market; it was a change in the composition of holders before it was a change in contractual fashion. The menu of contracts in existence is therefore an intersection rather than a design. A firm can issue only the allocations some holder is able and permitted to hold, and both words bind: a bond too small for the major indices is unbuyable in size by the funds that supply most of the demand, and a claim one notch below investment grade is unbuyable by an insurer whose capital charge steps at that boundary. Contracts whose control rights nobody is configured to exercise do not get written. Chapter 20's demand curves are accordingly not curves over payoffs alone; they are curves over control, and the price of a claim carries the cost of whatever balance sheet has to hold it. In Chapter 1 §1.2's terms, the price of a security reports a constraint binding at least as often as it reports news, and the contracts never written at all are that constraint binding where no price is printed to record it. *** ## Elsewhere in the Series * **The intellectual history** — the *History of Economic Thought* volume owns the development of institutional and new-institutional economics: the older American institutionalist tradition, Coase's route from the 1937 paper to the 1960 social-cost paper, Williamson's and North's Nobel-recognized programs, and the reception and critique of all of it. This chapter uses the ideas as working tools and does not narrate their history. * **This book.** Default as the state in which control transfers, and the pricing of that event: Chapter 10, especially §§10.1-10.2. Holder demand and the ecology that meets Part V's supply: Chapter 20, and Chapter 2's master holdings table for the claims themselves. Capital structure and payout as a division of cash-flow rights, with market timing and gap-filling: Chapter 23, especially §§23.4 and 23.7. Covenants, boards, compensation and activism as the fine structure of control: Chapter 24 §§24.1, 24.3. What firms actually issue, and the cross-country variation institutions produce: Chapter 25 §§25.1, 25.4. The takeover market as a governance mechanism: Chapter 22 §22.6. Equity as a residual claim priced as an option: Chapter 8 and Chapter 10 §10.2. *** ## Summary 1. **The Fisher Body episode is both the canonical hold-up story and a warning about canonical stories.** A 1919 ten-year exclusive-dealing contract at variable cost plus 17.6 percent, a boom in closed bodies, an alleged refusal to co-locate plants, and a 1926 acquisition — retold since Klein, Crawford and Alchian (1978) as hold-up defeated by integration. Coase (2000), Casadesus-Masanell and Spulber (2000) and Freeland (2000) disputed most of the factual core in a single symposium. Learn the mechanism; do not take the episode on faith. 2. **Coase asked why firms exist and answered with the costs of using the price mechanism.** Discovery, negotiation, and enforcement are costly, so some transactions are organized by direction rather than by contract. The firm is the supersession of the price mechanism. 3. **The boundary is a margin, not a direction.** A firm expands until the cost of organizing one more transaction internally equals the cost of transacting for it on the market or organizing it in another firm. Internal organization has rising costs of its own, which is why the economy is not one firm. 4. **The 1937 paper delivered a question with the right shape and a placeholder for the mechanism.** "Transaction costs" does not say which transactions are costly to contract for, and so generates no testable prediction until something fills it in. 5. **Williamson filled it in with asset specificity.** An investment worth much inside one relationship and little outside it produces the fundamental transformation: a large-numbers bidding situation becomes a small-numbers bargaining situation, created by the parties' own efficient investment. 6. **The cost of hold-up is underinvestment, not expropriation.** In Table 21.1, a supplier who captures half the gains from a specific investment costing 120 and worth 200 does not invest, and eighty of joint surplus is lost before any opportunism occurs. Long-term contracts, hostages, reciprocal exposure, and integration are all devices for raising the investor's share above the threshold. 7. **The empirical record is unusually supportive and unusually hard to construct.** Component-level automotive data, coal contract durations at mine-mouth plants, and the on-board-computer shock in trucking all line up with the prediction; specificity is nonetheless a constructed proxy, and causation is what the Fisher symposium fought over. 8. **Grossman, Hart and Moore explain what ownership is.** Contracts are incomplete because contingencies are unforeseeable, indescribable, or unverifiable. Residual control rights are the rights to decide what no contract covers, and ownership is the possession of them. 9. **Ownership should go to the party whose non-contractible investment matters most, and joint ownership is dominated.** In Table 21.2 the three feasible regimes deliver 380, 360, and 300 against a first best of 440. No regime attains the first best, because strengthening one party's incentive necessarily weakens the other's. 10. **A security is an allocation of cash-flow rights and control rights across states of the world.** Equity is residual cash flow plus residual control; debt is a fixed claim plus a rule that transfers control in default; covenants are the fine structure of the transfer; venture capital contracts unbundle the two dimensions explicitly. This is the bridge from the theory of the firm to every chapter that follows. 11. **Team production, the nexus of contracts, and institutions complete the frame.** Alchian and Demsetz make the monitor the residual claimant; Jensen and Meckling deny the firm any objectives of its own and supply the agency-cost apparatus that Chapters 23 and 24 deploy; North places contract enforceability inside a political and historical setting that Chapter 25 uses. 12. **Corporate finance opens here because the firm is a claim-issuing technology.** The boundary of the firm determines which cash flows are bundled into which claims; capital structure is a control allocation as well as a cash-flow division; governance is what residual control does. Part IV's holders and Part V's issuers meet over securities whose control content is part of what is being demanded and supplied. 13. **Which allocations get manufactured depends on which balance sheets can hold them.** Control rights collect where monitoring pays — a bank with a restructuring desk, a private credit fund, a venture partnership — while claims written for holders who cannot run a workout have their control content stripped out. A firm can issue only the allocations some holder is able and permitted to hold, which makes Chapter 20's demand curves curves over control as well as over payoffs, and makes the contracts that are never written a price signal with no price attached. *** ## Key Terms * **Transaction cost**: The cost of using the price mechanism — discovering prices, negotiating and writing contracts, and enforcing and adapting them. Positive transaction costs are the reason some transactions are organized inside firms * **Asset specificity**: The extent to which an investment supporting a transaction is worth more inside that relationship than in its next-best use; site, physical, human, and dedicated-asset forms * **Fundamental transformation**: The conversion of a large-numbers bidding situation into a small-numbers bargaining situation by the act of specific investment itself * **Hold-up**: The ex-post renegotiation of terms made possible by bilateral monopoly after specific investment is sunk; its characteristic cost is the underinvestment it induces ex ante * **Incomplete contract**: A contract that does not specify an action for every state of the world, because contingencies are unforeseeable, indescribable, or unverifiable by a court * **Residual control rights**: The rights to make decisions that no contract covers; under Grossman-Hart-Moore, ownership of an asset *is* the possession of these rights * **Nexus of contracts**: Jensen and Meckling's view of the firm as a legal fiction serving as the connecting point for contracting relationships among individuals, with no objectives of its own * **Make-or-buy**: The decision whether to produce an input internally or procure it on the market; the applied form of the boundary-of-the-firm question * **Governance structure**: The institutional arrangement within which a transaction is organized — market, hybrid (long-term contract, franchise, alliance), or unified ownership — assessed by its adaptive capacity and its bureaucratic cost * **Asset partitioning**: The separation of a firm's asset pool from its owners' personal estates in both directions; entity shielding is the direction contracts cannot cheaply replicate * **Default rules**: The terms corporate law supplies when the parties' own contract is silent — board authority, fiduciary standards, voting and quorum rules — which allocate residual control unless and until the charter contracts around them *** ## Readings ### Required * Coase, R. H. (1937). "The Nature of the Firm." *Economica* 4(16): 386-405. *Twenty pages, no mathematics, and the question that organizes this chapter. Read it for the margin at the end — the firm expands until internal and market organizing costs are equal — which is where the economics is.* * Grossman, S. J. and O. D. Hart (1986). "The Costs and Benefits of Ownership: A Theory of Vertical and Lateral Integration." *Journal of Political Economy* 94(4): 691-719. *The definition of ownership as residual control rights, and the result that integration is never a free improvement because it necessarily weakens the non-owner's incentives. The formal sections repay working through with Table 21.2 in hand.* ### Recommended * Williamson, O. E. (1979). "Transaction-Cost Economics: The Governance of Contractual Relations." *Journal of Law and Economics* 22(2): 233-261. *The clearest single statement of discriminating alignment: transaction attributes on one axis, governance structures on the other.* * Williamson, O. E. (2005). "The Economics of Governance." *American Economic Review* 95(2): 1-18. *A short late restatement of discriminating alignment by its author, and the practical substitute for the 1979 article in §21.2 for readers who will not sit with the original.* * Hart, O. and J. Moore (1990). "Property Rights and the Nature of the Firm." *Journal of Political Economy* 98(6): 1119-1158. *Extends Grossman-Hart to many assets and many agents, and delivers the results on complementary assets and on why joint ownership is generally dominated.* * Klein, B., R. G. Crawford and A. A. Alchian (1978). "Vertical Integration, Appropriable Rents, and the Competitive Contracting Process." *Journal of Law and Economics* 21(2): 297-326. *The source of the standard Fisher Body account and of "appropriable quasi-rent" as the object at stake.* * Coase, R. H. (2000). "The Acquisition of Fisher Body by General Motors." *Journal of Law and Economics* 43(1): 15-31. *The revisionist piece. It appeared in a symposium issue alongside Freeland's and Casadesus-Masanell and Spulber's reconsiderations and Klein's reply; read at least two of the four, and treat the exchange as a case study in what it takes to establish a fact about one firm.* * Jensen, M. C. and W. H. Meckling (1976). "Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure." *Journal of Financial Economics* 3(4): 305-360. *The nexus-of-contracts statement and the agency-cost taxonomy. Chapters 23 and 24 use the apparatus; read §§1-3 now and the rest with those chapters.* * Bolton, P., J. Scheinkman and W. Xiong (2006). "Executive Compensation and Short-Termist Behaviour in Speculative Markets." *Review of Economic Studies* 73(3): 577-610. *A reminder for §21.5 that the manager's own contract is one of the contracts in the nexus, and that its optimal design depends on what the firm's shares are doing — which makes the pay contract an object jointly owned by this chapter, Chapter 15, and Chapter 24 §24.3.* * Roumasset, J. (1978). "The New Institutional Economics and Agricultural Organization." *Philippine Economic Journal*. *Transaction and enforcement costs applied to sharecropping and tenancy rather than to make-or-buy. Useful in §21.5 precisely because the setting is so far from Fisher Body: the same first-best/second-best reasoning, and no vertical integration anywhere in it.* * North, D. C. (1991). "Institutions." *Journal of Economic Perspectives* 5(1): 97-112. *Twelve pages on institutions as the rules within which contracts get written and enforced. It is the bridge from §21.5's last paragraph to Chapter 25 §25.4's cross-country variation, and it is where the claim that enforceability is itself an outcome is stated most compactly.* * Hart, O. (2001). "Financial Contracting." *Journal of Economic Literature* 39(4): 1079-1100. *The single best companion to §21.4's bridge: it carries the incomplete-contracts argument from Grossman-Hart through Aghion-Bolton to security design in one continuous piece, and it is the survey to read if only one is read.* * Aghion, P. and P. Bolton (1992). "An Incomplete Contracts Approach to Financial Contracting." *Review of Economic Studies* 59(3): 473-494. *The formal version of §21.4's bridge: state-contingent control as the optimal financial contract, with control shifting to the investor in bad states.* * Kaplan, S. N. and P. Strömberg (2003). "Financial Contracting Theory Meets the Real World: An Empirical Analysis of Venture Capital Contracts." *Review of Economic Studies* 70(2): 281-315. *What Table 21.3's last row looks like in actual documents. The evidence that cash-flow, board, voting and liquidation rights are allocated separately and made contingent on performance is the closest thing the literature has to a direct test of the incomplete-contracts view of securities.* * Lafontaine, F. and M. Slade (2007). "Vertical Integration and Firm Boundaries: The Evidence." *Journal of Economic Literature* 45(3): 629-685. *The survey to read before believing any single vertical-integration study, including the ones cited in §21.2.* * Tirole, J. (2006). *The Theory of Corporate Finance*. Princeton University Press. *The standard graduate formalization of everything Part V treats verbally. Chapter 1 is the §21.6 companion; Chapter 3 up to §3.4 is the workhorse moral-hazard model that recurs throughout the book; Chapters 10-11 are the control-rights and takeover material. The exercise volume is the natural source of harder problems for this chapter and Chapter 24.* *** ## Discussion Questions 1. **What the Fisher Body controversy teaches.** The standard account and the revisionist accounts agree on the contract, the dates, and the acquisition, and disagree on nearly everything about motive and conduct. Suppose you were asked to settle it. List the specific evidence that would establish that Fisher held up General Motors, and for each item say who would have to have recorded it and why they might not have. Then generalize: which of the empirical claims you will meet in Chapters 22 through 25 — that acquirers overpay, that covenants bind, that firms have leverage targets — rest on evidence of the kind you have just found scarce, and which rest on evidence of a different kind? 2. **Does the firm's boundary matter to a diversified holder?** An investor holds the index of Chapter 17 and therefore owns both parties to most vertical relationships in the economy. If Fisher and General Motors are both in her portfolio, a transfer from one to the other leaves her wealth unchanged. Does the theory of the firm therefore have nothing to say to her? Work through the argument carefully, distinguishing transfers from deadweight losses, and connect your answer to Chapter 17's common-ownership discussion. Then state the condition under which the diversified holder should prefer integration and the condition under which she should be indifferent. 3. **When is a long-term contract as good as ownership?** Table 21.1 shows that anything raising the investor's bargaining share above the threshold restores efficient investment. If a contract can do it, why does anyone integrate? Give two situations in which a contractual solution should be expected to dominate and two in which it should not, and identify which feature of the transaction is doing the work each time. 4. **Ownership as a scarce instrument.** In Table 21.2 no ownership regime achieves the first best. Explain in your own words why not, without using the phrase "residual control rights." Then consider a firm with three critical assets and four managers whose investments all matter. What does the argument predict about the number of distinct owners, and what does it predict about how often those ownership assignments should change? 5. **What the boundary does to the securities that exist.** A conglomerate splits into two listed companies. From the standpoint of Part IV's holders, what has changed? Consider at least the following: the number of distinct claims, the diversification available inside each claim, the information disclosed, the covenant structures possible, and who can now be a marginal holder of each piece. Which of these effects would a Modigliani-Miller argument say is irrelevant to total value, and which does the argument of this chapter say is not? *** ## Problems **Problem 1 — Hold-up and underinvestment.** A supplier chooses a relationship-specific investment $$I \ge 0$$ at a cost of $$I$$. The investment generates a joint surplus from trade of $$\mathcal{S}(I) = 40\sqrt{I}$$, and the asset has no value outside the relationship. The supplier captures a share $$\phi$$ of the joint surplus in ex-post bargaining. (a) Find the first-best investment and the first-best net social surplus. (b) Find the supplier's chosen investment as a function of $$\phi$$, and evaluate it at $$\phi = 1/2$$. (c) Compute the net social surplus at $$\phi = 1/2$$ and state the efficiency loss. (d) The buyer offers to transfer ownership of the tooling, raising $$\phi$$ to $$3/4$$. Compute the new investment and the new net social surplus, and say how much of the efficiency loss the transfer recovers. (e) Show that the supplier's own net payoff is $$400\phi^2$$, and explain why the supplier's payoff and the social surplus are maximized at different values of $$\phi$$ only when the buyer also has an investment to make. **Problem 2 — Ownership allocation.** Two managers, F and G, each choose whether to make a non-contractible investment costing 130. F's investment raises the joint surplus of the relationship by 240; G's raises it by 220. The relationship generates 300 before either invests. One asset is critical; its owner captures three-quarters of the value his own investment creates, a non-owner one-half, and under joint ownership each party's veto reduces both to one-half. (a) For each of the three regimes — F owns, G owns, joint ownership — determine who invests and compute the total surplus net of investment costs. (b) Rank the regimes and compare with the first best. (c) How large would G's contribution have to be for G ownership to dominate F ownership, holding everything else fixed? State the answer as an inequality and explain the intuition in one sentence. (d) Now suppose the investment cost falls to 100 for both managers. Recompute part (a). What does the comparison say about when the ownership assignment matters most? **Problem 3 — Make or buy.** An automaker needs 200,000 units a year of a stamped body panel. (a) An outside supplier will deliver at 52 dollars per unit. Producing in-house requires a dedicated line costing three million dollars a year in fixed costs, plus 36 dollars per unit in variable cost. Compute the annual cost of each option and the annual volume at which the automaker is indifferent. (b) The panel is specific to one body design. The automaker estimates that renegotiation, expediting, and the adaptation delays that come with an outside supplier holding a specific asset cost it an extra amount per unit. How large must that amount be to flip the decision if internal production also carries 500,000 dollars a year of managerial attention and internal-influence costs that the market option does not? (c) Which of Coase's three categories of transaction cost does the per-unit adaptation cost in (b) belong to, and which category does the 500,000 dollars belong to? Explain why the second is the reason the answer to the boundary question is a margin rather than a direction. (d) The supplier offers a fifteen-year contract with an indexed price formula in exchange for the automaker funding half the tooling. Explain, in the terms of §21.2, what that contract is doing and what could still go wrong with it. Name the historical contract it resembles. **Problem 4 — Control rights in states of the world.** A firm reaches an interim date with debt of face value 80 outstanding, maturing at the final date. Liquidating now yields 70 with certainty. Continuing yields, at the final date, either a high or a low outcome with equal probability. All parties are risk neutral and there is no discounting. (a) *Case A*: continuing yields 150 or 20. Compute the expected total value of continuing, the expected payoff to debt, and the expected payoff to equity, under both continuation and liquidation. Which action is efficient? Which does equity prefer? Which does debt prefer? (b) *Case B*: continuing yields 90 or 20. Repeat the computations. Which action is efficient now, and which party's preference coincides with efficiency? (c) You may write a covenant that transfers the continuation decision to the creditor when a stated, verifiable condition holds. Design one that achieves the efficient action in both cases, and state exactly what the condition must be verifiable *about*. (d) Explain why no covenant can achieve the first best if the only verifiable variable is the firm's *past* cash flow rather than the distribution of its future one. Connect your answer to §21.3's definition of contractual incompleteness and to Chapter 24's treatment of covenant design. **Problem 5 — Discriminating alignment.** For each of the following, state the dominant transaction attribute, predict the governance structure Williamson's argument implies, and name one observable fact that would falsify your prediction: (a) an aluminum smelter and an adjacent alumina refinery; (b) a supermarket chain's purchases of bananas; (c) a pharmaceutical company's relationship with the university laboratory that discovered its lead compound; (d) an airline's purchase of jet fuel at a hub airport; (e) a film studio and a lead actor for a three-picture franchise. Then say which of the five you would expect the GHM framework to explain better than the transaction-cost framework, and why. *** ## Selected Solutions **Problem 1.** (a) Maximize $$40\sqrt{I} - I$$. The first-order condition is $$20/\sqrt{I} = 1$$, so $$\sqrt{I} = 20$$ and $$I = 400$$. Then $$\mathcal{S} = 40 \times 20 = 800$$ and the net social surplus is $$800 - 400 = 400$$. (b) The supplier maximizes $$\phi \times 40\sqrt{I} - I$$, giving $$20\phi/\sqrt{I} = 1$$, so $$\sqrt{I} = 20\phi$$ and $$I = 400\phi^2$$. At $$\phi = 1/2$$, $$I = 100$$. (c) At $$I = 100$$: $$\mathcal{S} = 40 \times 10 = 400$$, and the net social surplus is $$400 - 100 = 300$$. The efficiency loss is $$400 - 300 = 100$$, a quarter of the attainable total. Note that investment falls by three-quarters (from 400 to 100) while surplus falls by half and net surplus by a quarter — the concavity of $$\mathcal{S}$$ means the marginal units forgone are the least valuable ones, which is why hold-up is severe in investment terms before it is severe in welfare terms. (d) At $$\phi = 3/4$$: $$I = 400 \times 0.5625 = 225$$, $$\mathcal{S} = 40 \times 15 = 600$$, net social surplus $$600 - 225 = 375$$. The transfer recovers 75 of the 100 lost. (e) Substituting $$I = 400\phi^2$$ into the supplier's objective gives $$\phi \times 40 \times 20\phi - 400\phi^2 = 800\phi^2 - 400\phi^2 = 400\phi^2$$, increasing in $$\phi$$ throughout. The supplier always wants a larger share; social surplus $$800\phi - 400\phi^2$$ is also increasing on $$\[0,1]$$ and is maximized at $$\phi = 1$$. The two therefore agree here, and they agree only because the buyer makes no investment of his own. Give the buyer a non-contractible investment and $$\phi = 1$$ destroys it entirely, which is Table 21.2's result that ownership is a scarce instrument. **Problem 2.** (a) F as owner captures $$0.75 \times 240 = 180 > 130$$ and invests; as non-owner, $$0.50 \times 240 = 120 < 130$$ and does not. G as owner captures $$0.75 \times 220 = 165 > 130$$ and invests; as non-owner, $$0.50 \times 220 = 110 < 130$$ and does not. **Table 21.4: Ownership regimes in Problem 2** | Regime | F invests? | G invests? | Total surplus net of costs | | --------------- | ---------- | ---------- | ------------------------------- | | F owns | Yes | No | $$300 + 240 - 130 = 410$$ | | G owns | No | Yes | $$300 + 220 - 130 = 390$$ | | Joint ownership | No | No | $$300$$ | | First best | Yes | Yes | $$300 + 240 + 220 - 260 = 500$$ | *Source: Author's calculation, on the parameters stated in the problem.* (b) F ownership (410) beats G ownership (390), which beats joint ownership (300); all fall short of the first best of 500. Joint ownership is worst because a mutual veto makes both parties non-owners for incentive purposes. (c) Under either single-owner regime exactly one manager invests, so the comparison is between $$300 + 240 - 130$$ and $$300 + \Delta\_G - 130$$: G ownership dominates when $$\Delta\_G > 240$$. (Two side conditions must also hold for the regimes to look as described: $$0.75\Delta\_G \ge 130$$, so that G invests when he owns, and $$0.50 \times 240 < 130$$, so that F does not invest when he does not.) The intuition: with one asset to allocate, give it to whoever creates more value with it. (d) At a cost of 100: F as non-owner captures 120 > 100 and now invests, and G as non-owner captures 110 > 100 and now invests. Every regime, including joint ownership, delivers $$300 + 240 + 220 - 200 = 560$$, which is also the first best. The ownership assignment is irrelevant. The general point is that ownership matters most when investment incentives are marginal — when the non-owner's share is close to the threshold. Where investments are richly worthwhile regardless, or hopeless regardless, the boundary of the firm does no work. *** ## Data Exercise: Where Does Activity Happen? The chapter argues that the boundary of the firm is an economic margin. This exercise measures where that margin currently sits in the United States, using two free sources, and then asks a firm-level question with a licensed one. **Part A — The firm-size distribution (free data).** The Census Bureau's **Business Dynamics Statistics** (BDS) provides annual counts of firms, establishments, employment, and job flows by firm size and firm age, for the entire US private nonfarm economy. It is free, downloadable as flat files, and available by sector. 1. Download the BDS firm-size series for the most recent available year. Tabulate the number of firms, the number of establishments, and total employment by firm-size class. Compute each class's share of firms and share of employment. 2. Report the share of firms with 500 or more employees and their share of employment. The two numbers differ by more than an order of magnitude; state both and explain in one sentence what a Coasean reading of the gap would be. 3. Compute the average number of establishments per firm by size class. A firm with many establishments has chosen to organize geographically dispersed activity under one direction rather than through contracts among independent local businesses. Which size classes account for most of the multi-establishment employment? 4. Repeat step 1 for the earliest year in the series and compare. Has the share of employment inside large firms risen or fallen? Then state two measurement reasons why this series alone cannot tell you whether the *boundary of the firm* has moved: consider franchising and contract employment, both of which place activity outside the firm as BDS defines it while leaving it under substantial direction. **Part B — A make-versus-buy footprint (free data).** The Bureau of Economic Analysis publishes **input-output accounts** — the make and use tables — free, at several levels of industry detail. 1. From the summary-level use table, take three industries: motor vehicle manufacturing, computer and electronic products, and food services. For each, compute total intermediate inputs purchased from other industries as a share of total industry output. 2. Interpret that ratio as a crude "buy" share and one minus it as a crude "make plus value added" share. Then write down at least three reasons the interpretation is crude — start with the fact that value added includes labor compensation, which is bought on a market too, and that the industry classification aggregates over firms with very different boundaries. 3. For motor vehicles specifically, identify the largest single supplying industry in the use table and compare its share with what you would guess about vertical integration in the sector today. The industry that in 1926 absorbed its body maker now outsources a majority of component value; nothing in the table tells you why, but it tells you the magnitude of what has to be explained. 4. Optional: repeat step 1 for the earliest vintage of the use table available at a consistent industry definition, and describe the direction of change. Be explicit about which changes are real and which are reclassification. **Part C ★ (if you have WRDS).** Compustat's **segment files** report business and geographic segments for public firms, including segment sales and, where disclosed, intersegment sales. 1. Build a panel of firms reporting more than one business segment. Construct a diversification measure — the number of segments, or a sales-based Herfindahl across segments — and describe its distribution and its trend over the past three decades. 2. Using intersegment sales where reported, construct a firm-level vertical-relatedness measure: the share of one segment's output sold to another segment of the same firm. Report how many firms disclose enough to compute it, and treat the non-disclosure as data, not as a nuisance — what does the pattern of who discloses tell you? 3. Merge with an R\&D intensity measure and test whether vertically related firms are more R\&D intensive, which is the crudest possible proxy for the specificity of their investments. State the identification problem in one paragraph before reporting a coefficient, and say explicitly which direction of causation your design cannot rule out. 4. Finally, connect to Part V. For the firms in your sample that reported a segment divestiture or a spin-off, examine what happened to the number and type of securities outstanding afterward. That is §21.6's claim — that boundary decisions determine the menu of claims — stated as something you can count.