CHAPTER 4 Vertical Integration: Theory and Policy The law and economics of vertical integration have long been subject to controversy. Roger Blair and David Kaserman's recent review of the issues employs the language of warfarebattleground, skirmishes, campaigns, and the liketo set the stage (1983, p. 1). Contests of two kinds can be dis- tinguished. The earlier ones took place within the monopoly domain, the disputes having reference 1o whether vertical integration was principally an instrument of price discrimination, was designed to check successive margin- alization, or had entry barrier purposes. More recently monopoly and efficien- y explanations have been paired off. Vigorous resistance notwithstanding, the technological orientation and monopoly presumptions of an earlier era have gradually made way for an interpretation in which efficiency purposes are more prominently featured. By comparison with the 1968 Vertical Merger Guidelines, those issued by the Justice Department in 1982 make significant allowances for efficiency. Indeed, the 1984 Merger Guidelines even make Provision for an efficiency defense (Chapter 14). To be sure, as with most complex forms of organization, vertical integra- tion can and sometimes does serve a variety of economic purposes. i I focus here on what | consider to be the main purpose served: economizing on "Past Kicindorfer and Gunter Knicps (1942, p. 1) offer the following summary statement on & purposes. of vertical iniegration: s 102 / THE ECONOMIC INSTITUTIONS OF CAPITALISM 1982/1984 Vertical Merger Guidelines reflect a genuine sensitivity to transac- tion cost features-and are much more permissive than their predecessors as a consequence. Although there are those who counsel otherwise (Schwartz, 1983), public policy is arguably more informed and has been made more consonant with the public interest.86 / THE ECONOMIC INSTITUTIONS OF CAPITALISM Vertical Integration: Theory and Policy 1 87 transaction costs. A brief discussion of strategic purposes, however, is also realization of thermal economics is said to require integration (Bain, 1958, p. included. 381). Even, moreover, if tight technological linkages of that kind are missing, The commonsense explanation for vertical integration is that it has tech- existing configurations of assets are widely believed to reflect technological nological origins. That explanation is disputed in section 1. The main factor principles. Especially among noneconomists, more integration is thought to to which I attribute a decision to integrate is a condition of asset specificity. A be preferable to less. Only in such rare circumstances where outside suppliers simple model in which asset specificity is featured and in which the tradeoffs have patents or where economies of scale or scope are very large would between transaction costs and production costs are displayed is developed in outside procurement be seriously contemplated. section 2. Further implications of this approach are developed in section 3. All of the above is plausible, which is to say that vertical integration Vertical Merger Guidelines are examined in section 4. appears to be the unproblematic result of a natural technological order. I submit, however, that intermediate product market transactions are much 1. Technological Determinism more numerous than the conventional wisdom would suggest.' The marvels of the market to which Hayek referred in 1945 apply equally today. I further- Ours is indisputably a technologically advanced society. That complex orga- more contend that decisions to integrate are rarely due to technological deter- minism but are more often explained by the fact that integration is the source nization is needed to serve a complex technology is surely common sense. In of transaction cost economies. particular, comprehensive integration-backward into materials, laterally into components, and forward into distribution-is widely believed to be the One way of putting it is as follows: Technology is fully determinative of organizational means by which complex products and services are created, economic organization only if (1) there is a single technology that is de- cisively superior to all others and (2) that technology implies a unique organi- produced, and efficiently brought to market. zation form. Rarely, I submit, is there only a single feasible technology, and That conception is supported by the firm-as-production-function orienta- even more rarely is the choice among alternative organization forms deter- tion. Large, integrated firms, wherein production is accomplished by joining mined by technology. fungible inputs to yield outputs according to the engineering specifications, are supposedly the rule rather than the exception. Reference to "physical or Recall in this connection the contracting schema in Chapter 1, where technical aspects" sometimes buttresses this nonmarket presumption. The general purpose and special purpose technologies are distinguished. Recall standard example is the integration of iron and steel making, where the further that the parties to the transactions so described have the option of crafting governance structures responsive to their contracting needs. Only as market-mediated contracts break down are the transactions in question re- The most popular [explanation] has been that when economics of scope between successive moved from markets and organized internally. The presumption that "in the stages due to technological organizational interrelationships are strong enough, these ac- beginning there were markets" informs this perspective. livities should be provided under joint ownership (e.g.. Chandler [1966]). Other arguments This market-favoring premise has two advantages. One is that it helps to for Vertical Integration have been the avoidance of factor distortions in monopolized markets (e.g., Vernon and Graham [1971]. Warren-Boulion [1974]. Schmalensee [1973]): flag a condition of bureaucratic failure that has widespread economic impor- uncertainty in the supply of the upstream good with the consequent need for information by ance but goes little remarked. (The issues here are briefly introduced in downstream firms (Arrow [1975]); and the transfer of risks from one section of the econo- my to another (Crouhy [1976], Carlton [1979]). Furthermore, it has been pointed out that section 3 and are more fully developed in Chapter 6.) Second, it encourages transaction costs might create important incentives for vertical integration (e.g., Coase the view, which I believe to be correct, that technological separability be- [1937], Williamson [1971, 1975]). tween successive production stages is a widespread condition-that sepa- Omitted from this list is the incentive to use vertical integration as an organization shell to evade taxes on intermediate products (Stigler. 1951) or as a device, through judicious use of transfer 3Absent good measures of change in the amount of vertical integration, it is often inferred pricing. to take advantage of differences among tax jurisdictions (that arise, for example, between from the observed increases in firm size over time that the degree of vertical integration has states). Jerry Green's recent examination of information externalities (1984) also warrants inclu- increased. But such increases in firm size are often a result of radial expansion, whereby the firm sion. Yoram Barzel (1982) and Douglass North (1978) trace vertical integration to difficulties of grows to serve larger markets but the composition of activity is unchanged, or of diversification. measurement. It is evident that few consumer product firms are comprehensively integrated backward into raw materials. And many manufacturers decline to integrate forward into distribution. Lateral integra- The recent book by Roger Blair and David Kaserman (1983) provides an expansive survey tion into components is also incomplete--as an examination of the automobile industry (General and assessment of the literature. Motors, Ford, Chrysler, Toyota) discloses (Monteverde and Teece, 1982)88 / THE ECONOMIC INSTITUTIONS OF CAPITALISM Vertical Integration: Theory and Policy / 89 rability is the rule rather than the exception.".It thus becomes easy and even Note that reference to hazards introduces nonproduction (transaction natural to regard the transaction as the basic unit of analysis. As between cost) considerations. Of special importance in this connection, and the distinct alternative feasible modes for organizing transactions, which has superior contribution of transaction cost economics, is the following proposition: The efficiency properties and why? Once that orientation is adopted, internal magnitude of the hazards depends on the attributes of the assets and on the organization is seen less as a consequence of technology and more as the characteristics of the contracting relation. result of a comparative assessment of markets and hierarchies. Thus suppose that stage I requires an investment in durable, general A useful strategy for explicating the decision to integrate is to hold purpose equipment that is mounted "on wheels," hence can be costlessly technology constant across alternative modes of organization and to neutralize relocated. Contractual problems between independent buyer and supplier are obvious sources of differential economic benefit, such as transportation cost here limited since contracts can be terminated and productive resources relo- savings. Thus consider two separable manufacturing operations in which the cated at negligible cost. Given the unspecialized nature of the investments and output of one stage feeds the next. An entrepreneur has decided to enter stage the mobility that has been ascribed to them, neither buyer nor supplier oper- II activity and is considering alternative ways of organizing stage I. One ates at the sufferance of the other. Problems arise, however, if stage I involves possibility is to solicit bids from qualified suppliers to produce to his needs. A durable specialized investments or if, once put in place, relocation of general second is to integrate backward and do the work himself. purpose assets is thereafter very costly. Here the parties must face issues such Assume that the same stage I technology will be employed whether the as the following: Can the complex contract be written and implemented at low entrepreneur makes or buys. One factor that would appear to favor own- cost whereby independent parties assuredly adapt their relation efficiently to manufacture over procurement is that transportation cost economies may be changing circumstances? What are the hazards of incomplete contracting? In realized. That is superficial, however, since an independent stage I supplier consideration of the fundamental transformation to which autonomous con- can locate in the same cheek-by-jowl relation to stage II as can an integrated tracting is subject in these circumstances, ought unified ownership of the two owner. Accordingly, transportation (and related inventory cost savings) are stages be elected instead? Adaptive, sequential decision-making of the com- neutralized. What is it, then, that favors one of the modes in relation to the bined stages would then be implemented under the administrative aegis rather other? than in a recurring bargaining context." Although this query is one to which a theory of the firm might reasonably To be sure, this is a highly simplified and stylized example. But the basic be expected to speak, mundane vertical integration of this kind is a subject on argument applies quite generally: Technology is not determinative of eco- which the orthodox view of the firm as production function is curiously silent. nomic organization if alternative means of contracting can be described that Given that the two stages in question are technologically separable, and given can feasibly employ, in steady state respects at least, the same technology. I that factor price, tax, and related distortions are not obviously posed, there is submit that several alternative modes commonly qualify, whence technology no compelling neoclassical reason to prefer integration over market is more usefully regarded as a factor that delimits the set of feasible modes- procurement. the final choice thereafter turning on a transaction cost assessment. Dis- The notion that an independent stage I supplier would be willing to locate tinguishing among transactions according to their attributes is essential for in check-by-jowl proximity to the stage II buyer nevertheless runs contrary to final mode selection purposes. intuition. Surely there are undisclosed hazards in such an association? If so, Even that, however, is too simple. It assumes a sequential process does this have organizational ramifications? whereby technology is selected first and choice among feasible organizational modes is made thereafter. This convenient expository device is used in section "The view that technological nonseparability is a common condition and is primarily respons 2, below. In fact, however, technology and organizational modes ought to be sible for the appearance of the "classical capitalist firm" and its successor was advanced by treated symmetrically; they are decision variables whose values are deter- Armen Alchian and Harold Demsetz (1972). The principal example they offered in support of their view was that of manual freight loading. whereby two men need to work coordinately in mined simultaneously. The issues here are addressed elsewhere (Masten, order to load a truck. Such team relations are restricted to relatively small groups, however. The 1982; Riodan and Williamson, forthcoming). Suffice it to observe that, albeit symphony orchestra is the largest such group of which I am aware. See Karl Marx, Vol. 1. chap. 13, for an early insightful discussion of nonseparability. Interesting though such a condition is, "The limits of fiat also need to be addressed. Suffice it to observe here that common nonseparabilities do not explain the appearance and viability of very large-scale organization. ownership of successive stages attenuates incentives for managers to suboptimize. since they do Alchian (1984) now holds that complex organization owes its origin to transaction costs. "xx appropriate separate profit streams. But the issues are more complex. See Chapter 6.90 7 Tue Ecoxomic INSTITUTIONS OF CAPITALISM qualified, the main arguments survive when formulated in a more general framework. 2. A Heuristic Model As discussed in earlier chapters and sketched in section | above, the principal factor to which transaction cost economics appeals to explain vertical integra- tion is asset specificity. Without it, market contracting between successive production stages ordinarily has good economizing propertics. Not only can production economies be realized by an outside supplier who aggregates orders, but the governance costs of market procurcment are negligiblesince neither party has a transaction-specific interest in the continuity of the trade. As asset specificity increases, however, the balance shifts in favor of internal organization, The argument is developed in two parts. First, output is held constant and economies of scale and scope are assumed to be neghgible (or the firm in question is of sufficient size to exhaust them). Choice between firm and market thus turns entirely on govemance cost differences. Second, economies of scale and scope are admitted, but output is constrained to be the same. 2.1 Governance Costs and Economic Organization The main differences between market and intemnal organization are these: {1) Markets promote high-powered incentives and restrain bureaucratic distor- tions more effectively than internal organization; (2) markets can sometimes aggregate demands (o advantage, thereby to realize economies of scale and scope; and (3) internal organization has access to distinctive governance in- struments! The differences between market and internal organization in incen- tive and control respects are developed in Chapter 6. For my purpose here, 1 take these as given. ) Consider, therefore, the decision of a firm to make or buy a pa:r:icular good or service. Suppose that it is a component that is to be joined to the mainframe and that the quantity to be supplied is fixed.\" Economies of scale and scope are assumed to be negligible, so the critical factors that are deter- PAs | scknowledge above, a more complete and general model would treat outpa, asset specificity, and organization form as decision variahles. TAssume that the component is used in fixed propantions and represents a negligible fraction of the totsl cost, Vertical Integration: Theory and Policy ) ro9 Cost = FIGURE 4-1. Comparative Govemnance Cost minative in the decision to_ make or buy are production cost control and the ease of effecting intertemporal adaptation. ) T The high-powered incentives of markets manifest themselves in both respects: They favor tighter production cost control but, as the bilateral depen- dency of the relation between the parties builds up, they impede the case of adaptation. The latter effect is a consequence of the fundamental transforma- tion that occurs as a condition of asset specificity deepens. Let B (k) be the bureaucratic costs of internal governance and M(k) the comesponding gover- nance costs of markets, where k is an index of asset specificity. Assume that B(0) > M0}, by reason of the above described cost control effects. Assume further, however, that M' > B' evaluated at every k. This second is a conse- quence of the com.pmw disability of markets in adaptability respects. Letting &G = P(k) M(k), the relation shown in Figure 41 obtains. Thus market procurement is the preferred supply mode where asset spec- ificity is slight-because of the incentive and bureaucratic disabilities of internal organization in production cost control respects. But internal organi- zation is favored where asseq specificity is great, because a high degree of bilateral dependency exists in those circumstances and high-powered incen- tives impair the ease with which adaptive, sequential adjustments to distur- bances are accomplished. As shown, the switchover value, where the choice between firm and market is one of indifference, occurs at . 92 / THE ECONOMIC INSTITUTIONS OF CAPITALISM Vertical Integration: Theory and Policy / 93 2.2 Economies of Scale and Scope AC+ AG The foregoing assumes that economies of scale and scope are negligible, so hat the choice between firm and market rests entirely on the governance cost AC differences. Plainly that oversimplifies. Markets are often able to aggregate diverse demands, thereby to realize economies of scale and scope. According- ly, production cost differences also need to be taken into account. Again it will be convenient to hold output unchanged. Let AC be the steady state production cost difference between producing to one's own re- AG quirements and the steady state cost of procuring the same item in the market. (The steady state device avoids the need for adaptation.) Expressing AC as a function of asset specificity, it is plausible to assume that AC will be positive throughout but will be a decreasing function of k. The production cost penalty of using internal organization is large for standardized transactions for which market aggregation economies are great, whence AC is large where & is low. The cost disadvantage decreases but remains positive for intermediate degrees of asset specificity. Thus although dissimilarities among orders begin to appear, outside suppliers are nev- ertheless able to aggregate the diverse demands of many buyers and produce at lower costs than can a firm that produces to its own needs. As goods and services become very close to unique (k is high), however, aggregation econ- omies of outside supply can no longer be realized, whence AC asymptotically FIGURE 4-2. Comparative Production and Governance Costs approaches zero. Contracting out affords neither scale nor scope economies in those circumstances. The firm can produce without penalty to its own needs. This AC relation is shown in Figure 4-2. The object, of course, is not to "The argument assumes that the firm produces to and services only its own needs. If diseconomies of scale or scope are large, therefore, technological features will deter all but very minimize AC or AG taken separately but, given the optimal or specified level large firms from supplying to their own needs of asset specificity, to minimize the sum of production and governance cost Plausible though this appears. neither economies of scale nor scope are, by themselves. differences.. The vertical sum AG + AC is also displayed. The crossover responsible for decisions to buy rather than-make. Thus, suppose that economies of scale are large value of k for which the sum (AG + AC) becomes negative is shown by k, in relation to a firm's own needs. Absent prospective contracting problems, the firm could construct a plant of size sufficient to exhaust economies of scale and sell excess product to rivals which value exceeds k. Economies of scale and scope thus favor market and other interested buyers. Or suppose that economics of scope are realized by selling the final organization over a wider range of asset specificity values than would be good in conjunction with a variety of related items. The firm could integrate forward into observed if steady state production cost economies were absent. marketing and offer to sell its product together with related items on a parity basis-rival and complementary items being displayed, sold, and serviced without reference to strategic purposes. More generally, if * is the optimal degree of asset specificity," Figure That other firms, especially rivals, would be willing to proceed on this basis is surely 4-2 discloses doubtful. Rather than submit to the strategic hazards, some will decline to participate in such a scheme (Williamson, 1975, pp. 16-19, 1979c, pp. 979-80). The upshot is that all cost dif- 1. Market procurement has advantages in both scale economy and gov- ferences between internal and market procurement ultimately rest on transaction cost considera- crnance respects where optimal asset specificity is slight (4* > ). Not only does the market realize little ic distortions that appear in the unitary form (U-form) of enterprise. Ex- aggregate economy benefits, but market governance, because of the pressed in terms of Figure 4-2, the curve AG falls under multidivisionaliza- "lock-in" problems that arise when assets are highly specific, is tion as compared with the unitary form organization. Thus, assuming AC is hazardous. unchanged: 3. Only small cost differences appear for intermediate degrees of op- 6. An M-form firm will be more integrated than its U-form counterpart, timal asset specificity. Mixed governance, in which some firms will ceteris paribus. "! be observed to buy, others to make, and all express "dissatisfaction" with their current procurement solution, are apt to arise for these. Accidents of history may be determinative. Or nonstandard contracts 3. Further Implications of the types discussed briefly in Chapter 3 and examined more fully in Chapters 7 and 8 may arise to serve these. 4. More generally, it is noteworthy that, inasmuch as the firm is every- 3.1 Asset Specificity Distinctions where at a disadvantage to the market in production cost respects (AC > 0 everywhere), the firm will never integrate for production cost reasons alone. Only when contracting difficulties intrude does the Additional implications of a transaction cost economizing kind can be derived firm and market comparison support vertical integration-and then by recognizing that asset specificity takes a variety of forms and that the only for values of * that exceed k. organizational ramifications vary among these. Four types of asset specificity are usefully distinguished: site specificity-e.g. successive stations that are Additional implications may be gleaned by introducing quantity (or firm located in a cheek-by-jowl relation to each other so as to economize on size) and organization form effects. Thus consider firm size (output). The nventory and transportation expenses; physical asset specificity-e.g. spe- basic proposition here is that diseconomies associated with own-production cialized dies that are required to produce a component; human asset specifici- will be everywhere reduced as the quantity of the component to be supplied y that arises in a learning-by-doing fashion; and dedicated assets, which increases. The firm is simply better able to realize economies of scale as its represent a discrete investment in generalized (as contrasted with special own requirements become larger in relation to the size of the market. The purpose) production capacity that would not be made but for the prospect of curve AC thus everywhere falls as quantity increases. The question then is selling a significant amount of product to a specific customer. The organiza- what happens to the curve AG. If this twists about &, which is a plausible tional ramifications of each are as follows: construction, " then the vertical sum AG + AC will intersect the axis at a 1. Site specificity. Unified ownership is the preponderant response to value of k that progressively moves to the left as the quantity to be supplied an asset specificity condition that arises when successive stages are located in increases. Accordingly: close proximity to one another. Such specificity is explained by an asset immobility condition, which is to say that the setup and/or relocation costs are 5. Larger firms will be more integrated into components than will small- great. Once such assets are located, therefore, the parties are thereafter oper- er, celeris paribus. ating in a bilateral exchange relation for the useful life of the assets. Finally, although this anticipates arguments developed more fully in 2. Physical asset specificity. If assets are mobile and the specificity is Chapter 11, the bureaucratic disabilities to which internal organization is attributable to physical features, market procurement may still be feasible by concentrating the ownership of the specific assets (e.g. specialized dies) on subject vary with the internal structure of the firm. Multidivisionalization, the buyer and putting the business up for bid. Lock-in problems are avoided, because the buyer can reclaim the dies and reopen the bidding should Assume that P(t.X) = 1(k)X. where /(0) > 0 and /(k) is the internal governance cost per unit of effecting adaptations. Assume further that M(k,X) = M(k)X, where M(O) = 0 and M(k) is the corresponding governance cost per unit of effecting market adaptions. Then AG = [/() - "There are, however, offsetting considerations. U-form managers may be better able to M()JK, and the value at which AG goes to zero will be independent of X. The effect of implement their preferences for empire-building, which could take the form of vertical integra- increasing X is to twist AG clockwise about the value of & at which it goes to zero. tion. See Chapter 6.96 / THE ECONOMIC INSTITUTIONS OF CAPITALISM Vertical Integration: Theory and Policy contractual difficulties develop." Thus ex post competition is efficacious, procurement of items for which off-site production experiences little or no and internal organization is unneeded penalty. When is such a component bought, and when is it made? 3. Human asser specificity. Any condition that gives rise to substantial All these issues can be pulled together in the context of the "efficient human asset specificity-be it learning-by-doing or chronic problems of mov- boundaries" problem." Thus consider the organization of three distinct pro- ing human assets in team configurations-favors an employment relation duction stages, which, for site-specificity reasons, are all part of the same over autonomous contracting. Common ownership of successive stages is firm. That is the technological core. Suppose that raw materials are distinct thus predicted as the degree of human asset specificity deepens. and are naturally procured from the market. Suppose further that two things . Dedicated Assets. Investments in dedicated assets involve expand- occur at each production stage: There is a physical transformation, and com- ing existing plant on behalf of a particular buyer. Common ownership in these ponents are joined to the "mainframe. " And suppose, finally, that the firm circumstances is rarely contemplated. Trading hazards are nevertheless recog- nized and are often mitigated by expanding the contractual relation to effect has a choice between own distribution and market distribution. symmetrical exposure. Paradoxically, greater aggregate hazard exposure can Let the core production stages be represented by $1, $2. $3, and draw be mutually preferred to less if, as a consequence, hazard "equilibration" is these as rectangles. Let raw materials be represented by R and draw this as a thereby realized. (The issues here are developed more fully in Chapters 7 circle. Let component supply by represented by CI-B, C2-B, C3-B if the firm and 8.) buys its components, and CI-O, C2-0, C3-O if it makes its own components. Yet another implication of transaction cost reasoning is that where firms Draw these as triangles. Let distribution be given by D-B if the firm uses are observed both to make and to buy an identical good or service, the internal market distribution, and D-O if the firm uses own distribution. Draw these as technology will be characterized by greater asset specificity than will the squares. Finally, let a solid line between units represent an actual transaction external technology, ceteris paribus. ' No other approach to the study of and a dashed line a potential transaction, and draw the boundary of the firm as vertical integration generates this set of implications. a closed curve that includes those activities that the firm does for itself. Given the core technology presumptions, stages $1 through $3 will be organized internally and raw materials will be purchased. Components CI 3.2 Efficient Boundaries through C3 and stage D thus remain to be evaluated with respect to the tradeoffs set out in subsection 2.2 above. Assume that the firm determines on The foregoing treats every separable stage of production as one for which a this basis to make component C2 and engage in own distribution. The effi- careful assessment of make-or-buy is warranted. In fact, matters are often cient boundary of the firm is thus given by the closed curve in Figure 4-3 that simpler than that. There are some stages for which integration is not apt to be includes, in addition to the technical core, component C2 and the distribution seriously considered. Backward integration into raw materials is infeasible for stage, D. Components CI and C3 and raw material are procured in the many firms. Moreover, there are other stages for which common ownership will appear to be natural. James Thompson's references to a "core tech- market. Obviously this is arbitrary and merely illustrative. It also oversimplifies nology" (1967, pp. 19-23) presumes that some stages will be consolidated. greatly. It is relatively easy, however, to elaborate the schema to add to the Site specificity is commonly associated with these. More interesting is the 12See Tecce (1981) for an earlier discussion of this point. "4The term was first introduced by William Ouchi (1980a). One way of answering the question of whether drawing the boundary of the firm one way rather than another makes any Assume that the optimal level of & is very large, whichever mode of organization is difference is to ask a series of related questions. Consider the following: employed. Internal procurement is thus favored in these circumstances. Assume that this is done but that the firm thereafter opens a second plant (of identical size) in a different geographic region A. Production aspects for which it is constrained to procure the component in the market. The optimum level of & in 1. Would counomies of scale be mainly cahausted if the firm were to produce iss own requirements? hese two cases will not be identical. To the contrary, the choice of & in the second (noninte 2. Are economies of scope significant and can they be realized within the firm? grated) case will be less than in the first (integrated) case. The reason is that lower & will permit B. Design and asset aspects 1. Does the item in question have special design features? Should it? the market to realize aggregation economies and mitigate the governance costs of market procure- 2. Are steady state economies realized by producing the hem with the use of a special purpose ment. whereas internal organization is denied the same aggregation benefits (see, however, the technology! introductory remarks in Chapter 6) and has superior governance features. For a more general C. Contracting aspects discussion, see Riordan and Williamson (forthcoming). Also see Scott Masten (1982), who was 1. Are contracting parties prospectively locked into a bilateral exchange relation? he first rigorously to demonstrate this point. 2, Are there frequent needs to adapt the exchange relation to unanticipated disturbances?98 7 Tue ECONOMIC INSTITUTIONS OF CAPITALISM FIGURE 4-3. Efficient Boundary core, to consider additional components, to include several raw material stages and consider backward integration into them, to break down distribu- tion, and so on. But the central points would remain unchanged, namely: (1) The common ownership of some stationsthe coreis sufficiently obvious that a careful, comparative assessment is unneeded (site specificity will often characterize these transactions); (2) there is a second set of transactions in which own supply is manifestly uneconomic, hence market supply is indi- cated (many raw materials are commonly of this kind); but (3) there is 2 third set of activities for which make-or-buy decisions can be made only after assessing the production and transaction cost consequences of alternative modes. The efficient boundary is the inclusive set of core plus additional stages for which own supply can be shown to be the efficient choice. The basic orientation that informs the transaction cost approach to ver- tical integration is that integration should be selecrive. Contrary to what is sometimes argued, more integration is not always better than less. The data bear this out (see Chapter 5). 4. Vertical Merger Guidelines The cognitive map of contract set out in Chapter | (Figure 1-2) identifies two main approaches to the study of contract: monopoly and efficiency. OF the 'two, the monopoly approach was the more fully developed and more widely favored through the early 1970s (Coase, 1972). Vertical integration is one of the areas to which the monopoly branch of contract was thought to have relevance. Vertical Imtegration: Theory and Policy | 99 Leverage theory and entry barrier arguments were especially prominent, Vertical integration was believed to permit monapoly power in one area to be magnified through acquisition of another (the leverage !_'?9."'_'[.."_21"'_'!'_'.?\") or to impair the condition of eniry (the entry Barrier hypothesis). !> Lacking a *'physical or technical aspect'* m% whereupon tech- nological cost savings were plausibly associated with vertical integration, anticompetitive purpose was thought to be the driving force. It was easy, therefore, to conclude that public policy concern was warranted whenever vertical integration involved an *'appreciable degree of market control at even one stage of the production process (Stigler, 1955, p- 183). Specifically, Stigler stated that when s firm had at least 20 percent of an industry's outpu, its acquisition of more than 5 percent of the output capacity of firms to which it sells or from which it buys can be presumed to violate the antitrast laws (Stigler, 1955, pp. 183-84). T The 1968 Vertical Merger Guidelines, which set the limits on acquiring firm and acquired firm market shares at 10 and 6 percent respectively, were plainly in this monopolyftechnological spirit. The Guidelines were either informed by and reflected this line of scholarship, or the comespondence between the two is a remarkable coincidence. Given the prevailing firm-as- production-function framework, an affirmative rationale for vertical integra- tion that did have technological origins was not evident. There being no transaction cost economies to realize, even the slightest degree of monopoly power was thought to be responsible for decisions to integrate. The threshold level for imputing monoepoly power and purpase to the acquiring firm was sel al 20 percent by Stigler and was subsequently reduced to 10 percent in the 1968 Guidelines. Transaction cost economics takes issue with that in two respects. First, the possibility that vertical integration is driven by transaction cost economies needs to be taken into account where parties are operating in a bilateral trading context. Second, the slightest degree of monopoly power will not elicit inte- gration if, as developed more fully in Chapter 6, internal organization is beset by incentive difficulties. Slight degrees of monopoly power at one stage are not, without more, sufficient to warmrani internal procurement, This is not to say that vertical integration is wholly unproblematic in antitrust respects, however. To the contrary, integration by dominant firms can place smaller rivals at a strategic disadvantage. Interestingly, such anti- competitive effects also have transaction cost origins. VRobert Bork (1954) expressly took exception with this. but his views were not widely hoeded, 100 / THE ECONOMIC INSTITUTIONS OF CAPITALISM Vertical Integration: Theory and Policy / 101 Entry impediments of two types can arise where the leading firms in II firms where the HHI is below 1800. The 1982 Vertical Merger Guidelines stage I integrate (backward or forward) into what could otherwise be a com- hus focus exclusively on the monopolistic subset, which is congruent with petitively organized stage II activity. For one thing, the residual (noninte- transaction cost reasoning. It is furthermore noteworthy that the anticom- grated) sector of the market may be so reduced that only a few firms of petitive concerns to which the Guidelines refer-regarding costs of capital, 17 efficient size can service the stage II market. Firms that would otherwise be (contrived) scale diseconomies, and the use of vertical integration to evade prepared to enter stage I may be discouraged from coming in by the prospect rate regulation-are all consonant with transaction cost reasoning. 18 Also, the of having to engage in small-numbers bargaining, with all the hazards this 1982 Guidelines expressly acknowledge that investments in the secondary entails, with those few nonintegrated stage II firms. Additionally, if prospect market are risky in the degree to which "capital assets in the secondary live stage I entrants lack experience in stage II related activity, and thus would market are long-lived and specialized to that market. "19 This core proposition incur high capital costs were they to enter both stages themselves, integrated plainly has transaction cost economics origins. entry may be rendered unattractive. The integration of stages I and II by Despite this striking correspondence, the 1982 Guidelines are not fully leading firms is then anticompetitive. in entry aspects at least, if severing the consonant with transaction cost reasoning throughout. The transaction cost vertical connections would permit a competitive (large-numbers) stage II rationale for challenging a 5 percent acquisition whenever the HHI exceeds activity to develop without loss of scale economies. 16 1800 is not transparent. Furthermore, it was not until 1984 that the Guidelines Vertical integration in industries with low or moderate degrees of con- made provision for an economies defense. To be sure, there are hazards in centration does not, however, pose the same problems. Here a firm entering allowing an economies defense, especially if economic evidence must be into either stage can expect to strike competitive bargains with firms in the presented in court. " These hazards can be mitigated, however, if the Justice other stage whether they are integrated or nonintegrated. The reasons are that Department declines to bring cases where economies are clearly driving orga- no single integrated firm enjoys a strategic advantage with respect to such nizational outcomes (Kauper, 1983, pp. 519-22).?! transactions and that collusion by the collection of integrated firms (in supply Whatever is decided in economies defense respects, the fact is that the or demand respects) is difficult to effectuate. Vertical integration rarely poses an antitrust issue, therefore, unless the industry in question is highly concen- trated or, in less concentrated industries, collective refusals to deal are ob- 17To its credit, the Justice Department observes that the need for additional capital, by itself. does not constitute a barrier to entry into the primary market (here, stage 1), as long as the served. But for such circumstances, vertical integration is apt to be of the necessary funds are available at a cost commensurate with the level of risk in the secondary efficiency promoting kind. market. But the Department correctly recognizes that the risk in the secondary market is no The 1982 Merger Guidelines hold that vertical mergers are unlikely to independent of structure. Integrated entry that includes an unfamiliar stage is apt to carry a risk premium. This is because lenders "doubt that would-be entrants to the primary market have the pose troublesome antitrust issues unless the Herfindahl index in the acquired necessary skills and knowledge to succeed in the secondary market and, therefore, in the primary firm's market exceeds 1800 (this corresponds, roughly, to a four-firm con- market" (Guidelines, Sec. iv [B] [1] [b] [iD). The 1982 Guidelines further note that this problem centration ratio of 70 percent) and the market share of the acquired firm is exacerbated when a high percentage of the capital assets in the secondary market are long-lived and specialized to that market, and are therefore difficult to recover in the event of failure. exceeds 5 percent. The presumption is that nonintegrated stage I firms can Transaction cost reasoning plainly informs the 1982 reforms. satisfy their stage II requirements by negotiating competitive terms with stage The concern is that the regulator will be unable to evaluate the reasonableness of the costs incurred and prices charged by an integrated supplier because the relevant information is costly to obtain and difficult to evaluate. Such concerns would vanish were regulators comprehensively 16Although Posner now concedes that the above-described cost of capital effects can have knowledgeable (not subject to bounded rationality) or if regulated firms would disclose all entry barrier effects (1979, p. 946), he evidently regards this as a very special case. As Steven relevant information candidly (not subject to opportunism). Salop and David Scheffman point out, however, this is only one of a series of actions that dominant firms can take whereby differential cost penalities can be imposed on actual or potential 19Guidelines, See. iv (B) (1) (b). rivals (1983. p. 267). Among the tactics that they describe are (I) selective group boycotts (where 20Some of these are discussed in Williamson (1977). they offer the Klor's case as an illustration), (2) industrywide wage contracts, where wage 21 Refusal to consider compelling evidence that the parties to a transaction are proposing a increases have a disproportionate impact on the labor-intensive fringe (my treatment of the vertical merger because of the contracting difficulties that attend antonomous trading would Pennington cuse is an example; Williamson, 1968a). (3) vertical price squeezes, and (4) back- constitute a serious breach of rationality. General Motors' acquisition of Fisher Body after a ward integration by a dominant firm such that the downstream price to rivals of the affected input contracting relationship experienced strain is described by Klein, Crawford, and Alchian (1978). s raised differentially (Salop and Scheffman. 1983). Sce also Bain (1958, p. 658)