The discussion here does not pretend to be a comprehensive review of the prior literature that deals with vertical integration. It aspires merely to show that the vertical integration of technologically separable production stages ultimately turns on transactional considerations. Although this is sometimes suggested, it is rarely explicit in the literature; indeed, claims to the contrary are common.
1. Technological Interdependency
The technological interdependency argument is both the most familiar and the most straightforward: successive processes which, naturally, follow immediately in time and place dictate certain efficient manufacturing configurations; these, in turn, are believed to have common ownership implications. Such technical complementarity is probably more important in flow process operations ( chemicals, metals, and so forth) than in separable component manufacture. The standard example is the integration of iron- and steel-making, where thermal economies are said to be available through integration. It is commonly held that where “integration does not have this physical or technical aspect—as it does not, for example, in integrating the production of assorted components with the assembly of those components— the case for cost savings from integration is generally much less clear” (Bain, 1968, p. 381).
I submit, however, that, were it possible to write and enforce a complex contingent claims contract between blast furnace and rolling mill stages, the integration of these activities, for thermal economy reasons, would be unnecessary. The prohibitive cost of such contracting is what explains the decision to integrate. I furthermore contend, for the reasons developed in Section 4, that common ownership, by itself, is not sufficient to realize the potential economies of integration. The type of internal organizational structure that the firm employs also matters. Transaction costs thus both explain the decision to shift a transaction from the market into the firm and, within the firm, what organization form will be chosen.
Finally, I submit that even for transactions that do not involve “physical or technical” aspects of the kind referred to by Bain, integration may nevertheless permit transactional economies to be realized. Flow process economies between otherwise separable production stages are really a special case in which the small-numbers supply condition is particularly evident. Yet, absent thermal economies, or some counterpart physical condition, the opportunities to economize on transaction costs by resorting to vertical integration for exchanges that are conducted under conditions of uncertainty and small-numbers supply still obtain —albeit, perhaps, in less pronounced degree. Put differently, integration of the physical kind described by Bain is merely a special case of the contractual incompleteness issues addressed in Section 3.
2. Risk-Bearing and Moral Hazard
The moral hazard problem arises because of the conjoining of inhar- monious incentives with uncertainty—or, as Arrow puts it (1969, p. 55), it is due to the “confounding of risks and decisions.” To illustrate, consider the problem of contracting for an item whose final cost and/or performance is subject to uncertainty. One possibility is for the supplier to bear the uncertainty. But he will undertake a fixed price contract to deliver a specified result, the costs of which are highly uncertain, only after attaching a risk premium to the price. Assume that the buyer regards this premium as excessive and is prepared on this account to bear the risk himself. The risk can easily be shifted by offering a cost-plus contract. But this impairs the incentives of the supplier to achieve least-cost performance; the supplier may respond opportunistically by running slack (increasing on- the-job leisure) or reallocating his assets in such a way as to favor other work to the disadvantage of the cost-plus contract. (That these are more than mere hypothetical possibilities is revealed by the experience of military contracting on a cost-plus basis.)
Although, if commitments were self-enforcing it might often be institu- tionally most efficient to divide the functions of risk-bearing and contract execution (that is, cost-plus contracts would have ideal properties), specialization is discouraged by interest disparities. At a minimum, the buyer may insist on monitoring the supplier’s work. But he may also decide that integration is more attractive. Not only are the incentives to behave opportunistically thereby attenuated, but monitoring costs are apt to be less as well.
3. Variable Proportions Distortions
Consider the case where the assembly stage will support large numbers; fewness appears only in component supply. Whether monopolistic supply prices provide an occasion for vertical integration in these circumstances depends both on production technology and policing expense. Variable proportions at the assembly stage afford opportunities for nonintegrated assemblers to adapt against monopolistically priced components by substi- tuting competitively priced factors (McKenzie, 1951; Vernon and Graham, 1971, Schmalensee, 1973). Although the monopolistic component supplier might conceivably stipulate, as a condition of sale, that fixed proportions in assembly should prevail, the effectiveness of such stipulations is to be questioned – since, ordinarily, such promises will not be self-enforcing and information will be impacted to the disadvantage of the supplier, in which case the implied enforcement costs will be great. Where substitution occurs, inefficient factor proportions, with consequent losses to the system, will result. The private (and social) incentives to integrate, in order to reduce the total costs, by restoring efficient factor combinations, are evident.
It does not follow, however, that a net welfare gain will necessarily obtain. The effect on final price must also be considered. As Warren- Boulton has shown, forward integration into a competitive stage by a monopoly supplier can result in an increase in final price. Whether it will and, if it does, whether a net welfare gain is realized depends in a complex way on the elasticity of substitution, the elasticity of final demand, the condition of entry at the monopolized stage, and input cost relations (Warren-Boulton 1974).
It is commonplace that failure to make allowance for technological spillover between autonomous agents leads to suboptimization. While taxation is sometimes a feasible way to make such costs evident, integration may also be undertaken for this same reason (Davis and Whinston 1961).
An alternative suggested by Coase (1960) for dealing with small- numbers externalities is bilateral negotiation. Assuming bargaining costs to be negligible, the assignment of property rights with respect to spillovers to one party or the other can be shown to have no allocative efficiency con- sequences.47 But the zero bargaining cost fiction is plainly unrealistic.
As between merger and autonomous market contracting, the question is which mode incurs the lowest transaction costs. In static markets (see Section 2), the two modes would presumably be regarded with indifference. With the introduction of uncertainty, however, successive adaptations to changing market circumstances are needed. A once-and-for-all merger agreement is here apt to entail lower costs than recurrent bilateral bargaining — whence the occasion to integrate again turns on transaction cost considerations.
5. Markets for Information
5.1. OBSERVATIONAL ECONOMIES
As noted in the discussion of peer groups in Chapter 3, “the acquisition of information often involves a ‘set-up cost’; that is, the resources needed to obtain the information may be independent of the scale of the production process in which the information is used” (Radner, 1970, p. 457). Although Radner apparently had horizontal firm-size implications in mind, the argu- ment also has relevance for vertical integration. If a single set of observations can be made that is of relevance to a related series of production stages, vertical integration may be efficient.
Still, the question might be raised, why common ownership? Why not an independent observational agency that sells information to all comers? Or, if the needed information is highly specialized, why not a joint venture? Alternatively, what inhibits efficient information-exchange between successive stages of production according to contract? In relation, certainly, to the range of intermediate options potentially available, common ownership appears to be an extreme response. What are the factors which favor this outcome?
One of the problems with contracts is that of specifying terms. But even if terms could be reached, there is still a problem of policing the agreement.
To illustrate, suppose that the common information-collection responsibil- ities are assigned by contract to one of the parties. The purchasing party en runs a veracity risk: information may be filtered and possibly distorted to the advantage of the firm that has assumed the information-collection responsibility. If checks are costly and proof of contractual violation dif- ficult, contractual sharing arrangements manifestly experience short- run limitations. If, in addition, small-numbers conditions prevail, so that options are restricted, contractual sharing is subject to long-run risks as well.
n this argument integration for purposes of observational economies is again to be traced ultimately to transaction-cost considerations.
5.2. CONVERGENCE OF EXPECTATIONS
The issue to which the convergence of expectations argument is addressed is that, if there is a high degree of interdependence among suc- cessive stages of production and if occasions for adaptation are unpredic- table yet common, coordinated responses may be difficult to secure if the separate stages are operated independently. March and Simon (1958, p. 159) characterize the problem in the following terms:
Interdependence by itself does not cause difficulty if the pattern of interdepen- dence is stable and fixed. For in this case, each subprogram can be designed to take account of all the subprograms with which it interacts. Difficulties arise only if program execution rests on contingencies that cannot be predicted perfectly in advance. In this case, coordinating activity is required to secure agreement about the estimates that will be used as the bases for action, or to provide information to each subprogram unit about the activities of the others.
This reduces, in some respects, to a contractual incompleteness argu- ment. Were it feasible exhaustively to stipulate the appropriate conditional responses, coordination could proceed by contract. This is ambitious, however; in the face of a highly variable and uncertain environment, the attempt to program responses is apt to be inefficient. To the extent that an unprogrammed (adaptive, sequential) decision-process is employed instead, and in consideration of the severe incentive and control limitations that longterm contracts experience in these circumstances, vertical integration may be indicated.
But what of the possibility of short-term contracts? It is here that the convergence of expectations argument is of special importance. Thus, assume that short-term contracts are defective neither on account of investment disincentives nor first-mover advantages.49 It is Malmgren’s (1961) contention that such contracts may nevertheless be vitiated by the absence of structural constraints: the costs of negotiations and the time required to bring the system into adjustment by exclusive reliance on market (price) signals are apt to be great in relation to that which would obtain if successive stages were integrated and administrative processes employed as well or instead.
5.3. CAPITAL MARKETS
Richardson poses the problem of interest here by reference to an entre- preneur who was willing to proffer long-term contracts (at normal rates of return, presumably) that others were unprepared to accept because they were not convinced that he had “the ability, as well as the will, to fulfill them. He may have information sufficient to convince himself that this is the case, but others may not” (Richardson, 1960, p. 83). He goes on to observe that the perceived risks of the two parties may be such as to make it difficult to negotiate a contract that offers acceptable risks to each; objective risks are augmented by contractual risks in these circumstances. Integration undertaken for this reason is akin to self-insurance by individuals who know themselves to be good risks but are priced out of the insurance market because of their inability, at low cost, to “reveal” this condition to insurers. (See the discussion of insurance in Section 3, Chapter 1.)
The problem is not that investors are “uninterested” in financing projects of this sort. Rather, it is their inability to discriminate among good risks and poor, coupled with the tendency for poor risks to exaggerate their qualifications, that gives rise to the difficulties. The pairing of information impactedness with opportunism thus again supplies a selective incentive to integrate.51
A somewhat special, but nevertheless interesting, occasion to substitute vertical integration for autonomous contracting comes up in conjunction with the management succession problem in service industries. Investment bankers are, or at least have been, apprehensive about taking such businesses public. This may just be partly a matter of unfamiliarity with nonmanufacturing businesses. But bankers are also reluctant to invest in human assets. Unlike the manufacturing business, where a considerable fraction of the value of the firm is represented by durable physical assets, the principal assets of value in many service industries are the human assets of a going concern. The latter can be dissipated relatively quickly, before bankers can be apprised and take remedial action.
Faced with such capital market difficulties, many service firms, when confronted with a management succession problem, turn to a manufacturer with which they have already had business dealings to attempt a merger instead. The advantages that the manufacturer has over the banker are ( ) is pre-existing familiarity with the business, and ( 2) the greater control t at results from merger, as compared with lending, over both strategic decision- making and operating affairs in the acquired enterprise (see Chapters 6 and 7).
Source: Williamson Oliver E. (1975), Markets and hierarchies: Analysis and antitrust implications, A Study in the Economics of Internal Organization, The Free Press.