The Limits of Firms: The Costs of Bureaucracy

The costs of acquisition discussed in section 2 are mainly ones that will accrue to any separation of ownership from control, merger-related or otherwise. Although the costs of acquisition discussed in section 3 are not confounded in ownership and control respects, they are also more speculative in nature and are probably weaker in effect. The question thus arises as to whether there are other costs of merger that have yet to be identified. In particular, are there undisclosed “costs of bureaucracy” that obtain when successive production stages are joined?

Philip Selznick contends that “the most important thing about [nonmarket] organizations is that, though they are tools, each nevertheless has a life of its own” (1949, p. 10). Instrumental intentions notwithstanding, formal structures can “never succeed in conquering the nonrational dimensions of organizational behavior” (Selznick, 1948, p. 25). Richard Scott summarizes the argument as follows:

[O]rganizationa! rationality is constrained by “the recalcitrance of the tools of action”: persons bring certain characteristics to the organization and develop other commitments as members that restrict their capacity for rational action; organizational procedures become valued ends in themselves; the organization strikes bargains with its environment that compromise present objectives and limit future possibilities. [1981, p. 91]

What are the ramifications of such views for economic approaches to the study of organization? One possible economic response is to regard the condi- tions to which Selznick, Scott, and others refer as noise. Aberrations from rationality are thus treated as error terms. A stronger response is to deny that such behavior even exists. Neither is proposed here. Instead, an informed economic approach to the study of organization will display an interest in and thereafter make provision for all regularities of whatever kind. If the behavior in question is systematic, allowance can be made for it in making comparative institutional choices and in organizational design respects. Thus, if some forms of organization are less subject to bureaucratic distortions than others, this will be taken into account in assessing alternative modes. Within modes where distortions are especially severe, moreover, it may be possible to mitigate such conditions by devising checks or organizational reforms.

By comparison with the market failure literature, the literature on bu- reaucratic failure is relatively underdeveloped. The discussion here merely attempts to identify some of the main life cycle features that beset internal organization. As compared with market organization, internal organization displays a differential propensity to manage complexity, to forgive error, and to engage in logrolling.

1. The Propensity to Manage

A propensity to manage seems to characterize all forms of bureaucratic orga-nization. To be sure, the public sector is widely thought to be especially culpable in this respect—and probably is. Charles Morris captures the spirit in his reference to the cost of good intentions”. What he characterizes as ‘‘the new rationalist style in government” was based on “a confident optimism that the most intractable problems would give way before the resolute assault of intelligent, committed people” (1980. p. 23). But the same attitude also characterizes the private sector (Feldman and March, 1981).

Actually, the propensity to manage has two parts rather than just one. The part to which Morris refers is the instrumental propensity: Decisionmakers project a capacity to manage complexity that is repeatedly refuted by events. Although such a propensity is well-intentioned, problems regularly turn out to be more difficult and/or managerial competence more limited than managers of complexity originally project (Perrow, 1983). Counterexamples to the contrary notwithstanding, this is the main opinion.

The second type of propensity is more reprehensible. It is the strategic propensity to use the resources of the organization to pursue subgoals. If, for the reasons given in section 2, pecuniary incentives in firms are weaker than those in markets, then political games and preferences have greater sway. Efforts to tilt the organization, often through greater hands-on management commonly result.

Odysseus-type solutions are often attractive where ex ante resolve is known- regularly to break down in the face of recurrent exigen- cies/temptations. Out of awareness that the call of the Sirens would be well- nigh irresistable, Odysseus instructed that he should be bound to the mast. As Jon Elster points out, binding oneself is a privileged way of resolving the problem of weakness of will” (1979, p. 37). Such self-denial benefits can be rea lze wit in the firm if mergers for which the advocates can identify only limited prospective benefits are refused. The uncounted, but nevertheless predictable, costs attributable to the future propensity to manage will thereby assuredly be avoided.

2. Forgiveness 

Systems of justice vary systematically with organization. Families are ordinarily presumed to have deep knowledge of the transactions that occur between or impinge on the membership, to employ long time horizons, and to be relatively forgiving. Markets, in contrast, are presumed to have less knowledge of idiosyncratic circumstances, employ shorter time horizons, and are relatively severe (unforgiving).

For the reasons given in section 5, internal organization enjoys comparative auditing advantages. Accordingly, the integrated firm has greater capacity to reach informed decisions on the merits than do nonintegrated trading entities. The confounding of risks and decisions, which complicates market assessments, can thus be unpacked with greater precision and confidence internally. In principal, therefore, internal asset managers can better ascertain whether to continue funding a project than could the capital market.

But there are at least two further consequences. First, the prospect of penalty can and often does elicit inordinate energies. The market is a severe taskmaster. Unless escalator clauses have been expressly agreed to from the outset, unexpected cost increases are absorbed rather than passed through when transactions are mediated by the market. By contrast, unexpected cost increases that occur in trades within the firm are apt to be negotiable. The justification for such increases can be examined more fully within the firm, and the hazards of misrepresentation are less severe than in the market. But internal organization is thereby denied access to the supranormal energies that the market is able to mobilize. It is unrealistic to expect that such efforts will be expended if reasoned or plausible explanations can be advanced to support the cost increases in question: Reasoning systems are expected to behave in reasoning ways. (Academics, being the ultimate reasonersr are often unsuited for administrative positions on that account.)

Second, the net benefit calculus employed by firm and market differ. Indeed, a useful definition of forgiveness, at least for purposes of evaluating commercial transactions, is whether “excuses” are evaluated strictly with reference to a pecuniary net benefit calculus or not. As between the two, the market is expected to employ a stricter pecuniary net benefit calculus than is the firm. In this sense, it is less forgiving. A leading reason for this is that the firm maintains greater separability among transactions with the market than it can for transactions that are organized internally.

It is thus easy for a firm to terminate a supplier of a good or service that is supported by nonspecific assets (k = 0) at the first indication of failure. Were this same transaction to be organized internally, however, the firm would conduct an inquiry and consider second chances. Partly this is a manifestation of the aforementioned auditing advantage of internal organization. But it is also the case that the firm is unable to treat each internal transaction in a fully separable way. In effect, internal transactions of the k = 0 kind benefit from an association with other internal transactions for which k > 0.

Thus whereas continuity of transactions of the latter kind is valued from a pecuniary net benefit standpoint, the same is not true for the former. Firms that internalize transactions of both kinds are unable, however, to treat each in a fully discriminating way. A rational decision to “work things out” when things go wrong for k > 0 transactions spills over and infects transactions of the k = 0 kind. It is unacceptable, both to insiders and to interested observers, for the firm to behave differently. Simple regard for human dignity demands that due process be respected. Barnard’s remarks about informal organization are apposite: One of the purposes served by informal organization is that of “maintaining the personality of the individual against certain effects of formal organizations which tend to disintegrate the personality” (Barnard 1938 p. 122).

Thus whereas extreme market outcomes can be accepted as the luck of the draw, administrative actions are interpreted by all of the affected parties, including interested outsiders (associates, family, friends, and so forth), as merit choices. That places a severe burden of due process on internal organization. A plausible case will not do; a preponderance of evidence is needed if severe penalties are to be meted out. Out of awareness that they operate under the protection of a norm of internal due process, individuals are able to exploit the internal organization in minimum performance respects, and some do.

To be sure, the above described weakening of incentives applies strictly to k = 0 activities. The possibility that a merit choice environment intensifies incentives within the firm, as compared with the market, for k > 0 activities is not denied. It suffices, however, for the purposes of the argument, that due process spillovers are responsible for incentive attentuation for the k = 0 condition.

3. Logrolling 

Again the issue is not whether internal organization experiences costs but whether differential costs are incurred in moving from nonintegrated to inte- grated status. I submit that internal operating and investment decisions are more subject to politicization.


Alvin Gouldner contends that the norm of reciprocity is as important and universal as the incest taboo (1961). It finds expression throughout human society—across cultures and degrees of development and over time. Oppor- tunities to give effect to reciprocity, however, vary with the circumstances. In general those opportunities are greater within more highly integrated organi- zations than in less. Moreover, those opportunities are apt to be given ex- pression—which is a manifestation of the earlier described tendency for internal traders to be more accommodating than autonomous traders. This is not necessarily a bad thing; the prospect of such accommodation is a factor that weighs favorably in evaluating trades that are supported by specific assets. The possibility that it will go beyond the objective merits to include reciprocal managerial back-scratching is, however, a matter for concern and is appropriately taken into account in the decision to internalize incremental transactions.


An internal procurement bias is supported by a number of factors.15 For one thing, the internal supplier that produces mainly for internal uses may be judged to be at a relative disadvantage in the market place. The internal supplier may lack both the large and experienced marketing organization and the established customer connections to which nonintegrated external suppliers have access. In consideration of such conditions, and if fixed costs are nonredeploy able, a “preference” for internal procurement might seem appropriate—at least so long as the external price exceeds the variable cost of internal supply.

This may be a specious argument, however, since the nonredeployability of assets may easily be overstated (there may be a secondhand market for the machinery in question) and individual equipment renewal decisions ought eventually to be made with reference to the long-run viability of the internal facility. Managers are notably reluctant, however, to abolish their own jobs, even in the face of employment guarantees. The problems with such guarantees are that while continued employment may be secure, assurances that status will be maintained when a position is eliminated, and that promotion prospects will not be upset upon removal from a known promotion ladder are unenforceable. A preference for internal supply is thus to be expected and may manifest itself by urging that each equipment replacement decision be made serially, in semi- independent fashion. A fundamentally nonviable internal capability may be uncritically preserved in this way.

4. Further Remarks 

The main benefits of vertical integration, according to the transaction cost economics point of view adopted here, take the form of governance rather than production cost savings. They are discerned by examining the problems that attend autonomous contracting when the parties to a trade are operating in a bilateral exchange relation. The main costs of vertical integration are more difficult to discover, however. They are plainly not of a neoclassical production function kind. Neither are differential governance cost features transparent. Analysis at a more microanalytic level is evidently needed.

That is an analytic inconvenience, to say the least—made all the more so by the underdeveloped state of the bureaucratic failure literature. One pos- sibility, when the economic realities do not line up with the analytic conve- niences, is to pretend otherwise. Just as transaction cost reasoning and the examination of microanalytic phenomena has helped to illuminate the factors responsible for market failure, however, I conjecture that similar headway can be made against the subject of bureaucracy if a concerted effort is made here as well.

To be sure, undiscovered production cost features may appear that vitiate the need for such an effort. Or unremarked contractual difficulties that attend vertical integration may still emerge. I will be surprised, however, if the principal limits to vertical integration turn out to have nonbureaucratic origins. A difficult road ahead is in prospect.

Source: Williamson Oliver E. (1998), The Economic Institutions of Capitalism, Free Press; Illustrated edition.

Leave a Reply

Your email address will not be published. Required fields are marked *