Vertical integration, and the evidence that relates thereto, of two kinds is usefully distinguished. The first, which 1 will refer to as mundane vertical integration, involves integration of successive stages within the core technology. (These are the on-site stages referred to as SI, S2, and S3 in Figure 4-3.) The second, which is more exotic, involves integration of peripheral or off-site activities—backward integration into basic materials, lateral integration into components, forward integration into distribution, and the like’. Most discussions of vertical integration pass over the first and focus entirely on the second. But for the present brief discussion, this chapter also follows this practice.
It bears repeating, however, that integration of the core technology— stages that are located in a cheek-by-jowl association with one another, thereby to save on transportation and inventory expense, realize thermal economies, and the like—is not so unproblematic that this should be taken for granted. It also bears remark that the orthodox theory of the firm has no explanation for why successive stages in the core technology should be under unified ownership rather than each owned autonomously.
Inasmuch as there is overwhelming evidence that successive cheek-by- jowl stages are integrated, failure for a theory of the firm to explain this condition constitutes a serious lapse. Theories which do, by contrast, offer consistent explanations for both on-site and off-site types of integration are presumably to be credited (or, as the case may be, discredited) with evidence of both kinds.
The evidence that successive on-site stations are predominantly .integrated is abundant. What is everywhere taken for granted is not, however, beyond review. The integration of flow process operations is especially thought to be obvious. Consider, for example, the ownership and operation of successive stages within a petroleum refinery.
Although it is common to assume that a refinery is an indecomposable technological unit, whence the interesting questions of integration involve assessments of backward vertical integration into crude oil supply or forward integration into distribution of refined product, this technological presumption is incorrect. Numerous separable stages within the petroleum refinery can be identified, the organization of which is problematic. How should the storage tanks for intermediate and finished product be owned and operated? Should the asphalt unit be franchised to the highest bidder? Should the quality control laboratory be independently owned and operated? Such queries are rarely posed, but they are plainly matters to which a theory of economic organization might reasonably be asked to speak.
One of the reasons, I submit, why these mundane matters go unremarked is because most of us have reasonably good transaction cost intuitions. Nev- ertheless, transaction cost economics asks that these intuitions be probed by examining the attributes of transactions, with special emphasis on the condition of asset specificity. Potentially troublesome transactions are ones where the parties are effectively operating in a bilateral exchange relation to each other and need to adapt the interface at recurrent intervals. These are precisely the circumstances where asset specificity, uncertainty, and frequency are joined.
The storage tank and asphalt unit stages described above are almost certainly characterized by a high degree of site specificity. Recovery value that exceeds scrap is unlikely for the storage tanks; and the asphalt unit can be redeployed to alternative use or users only at grèat expense. The redeployability of the quality control laboratory might conceivably be preserved by siting it, albeit at added expense, on wheels. In the degree to which idiosyncratic knowledge of the refinery has quality control importance, however, a human asset specificity condition intrutjes.
Vertical integration is thus the predicted response for both storage tank and asphalt units. Unless the physical and human assets of the laboratory can be moved at slight sacrifice, moreover, integration is the preponderant response to the issue of laboratory organization as well. Transaction cost economics further predicts that if, for regulatory or other reasons, prohibitions or penalties against vertical integration for these transactions are posed, then long-term contracts will be devised in which bilateral (private ordering) safeguards are carefully crafted (Joskow, 1985).
Suppose, arguendo, that the capacity of transaction cost economics to reach and deal with mundane integration is granted. Forward, lateral, and backward integration are surely more problematic, however. How does it fare in these respects? Consider these seriatim.
Source: Williamson Oliver E. (1998), The Economic Institutions of Capitalism, Free Press; Illustrated edition.