1、11 Managerial Discretion,Organization Form,and the Multi-division Hypothesis!Oliver E.Williamson Implicit in my previous research on the sources and consequences of managerial discretion 43,46 has been the assumption that the firm is organized along the lines of what I will refer to here as the unit
2、ary form.Although the possibility of divisionalization was recognized and some of the organizational effects of divisionalization were briefly discussed(43,pp.120,124,160-1),these were left mainly undeveloped.I attempt here to remedy this by focusing explicitly on the multidivision structure.What,in
3、 relation to the unitary form,are its structural and performance attributes?Although one might attempt instead to determine the properties of the multi-division form by evaluating it with respect to a hypothetical ideal,this comparative approach would appear to have two advantages.First,it exposes d
4、imensions of the corporate control problem that may otherwise fail to be invoked.Second,much of what will prove testable in the theory of the firm is apt to relate not to some ideal standard but to the comparative performance of alternative organization forms.Lest I be misunderstood,I should make it
5、 clear that the emphasis is on the central behavioral tendencies and the differential capacities for superior performance that these two organization forms possess.Individual comparisons are therefore hazardous,but on the average 1 Research on this paper was supported in part by grants from the Nati
6、onal Science Foundation and the Brookings Institution.The views expressed are my own.This paper is based on research that is developed more extensively in 49.Helpful comments from David Conrath,Dale Henderson,Louis Pondy,Richard Posner,Sidney Winter,Jr,and Adrian Wood are gratefully acknowledged.Mor
7、e generally,the discussion of these issues by the participants at the September 1969 Glen Cove Conference,where the paper was first presented,is appreciated.344 OLIVER E.WILLIAMSON(assuming that the argument is correct)the differences indicated should obtain.The paper begins with a brief examination
8、 of the structure and performance of the unitary-form organization,including a review of conventional competition in the capital market arguments.Section II introduces the multi-division form innovation.The unitary and multidivision structures are compared and the multi-division hypothesis is advanc
9、ed in section III.The efficacy of competition in the capital market is re-examined in the light of this discussion in section IV;the implications of the argument for the conglomerate form of organization are also developed in this connection.Certain qualifications to the argument are given in sectio
10、n V.Some concluding remarks concerning organizational innovation and institutional failures are offered in section VI.For convenience in the exposition,unitary and multi-division enterprises will be referred to as U-form and M-form organizations respectively.I.THE UNITARY-FORM ENTERPRISE The discuss
11、ion of the unitary or U-form enterprise is in two parts.First,the internal organization and resulting performance characteristics of the U-form firm are examined with respect to increases in firm size.Second,the extent to which the capital market serves as an effective constraint on discretionary be
12、havior in an environment dominated by U-form firms is considered.1.Organization of the U-form enterprise The principal operating units in the U-form enterprise are the functional divisions-sales,finance,manufacturing,etc.Specialization by function has been the traditional way of organizing multifunc
13、tional activities and,in an enterprise of only moderate size,is the prevailing basis today.An organization chart displaying this structure is shown in Fig.11.1.The U-form organization can reasonably be represented as one which,in firms of moderate size,solves the division of labor question efficient
14、ly-including the provision for an effective strategic decision-THE MULTI-DIVISION HYPOTHESIS 345 making and control apparatus.Our interest here,however,is in firms of large and particularly giant size.Arbitrarily this large-firm universe will be taken to be the 500 largest industrials,50 largest uti
15、lities,and 50 largest transportation companies as ranked in the Fortune 500 series in a recent year.The giant-firm subset will be taken to be the 25 largest industrials,the 10 largest utilities,and the 10 largest transportation companies.Of the roughly 40 per cent of national income and employment t
16、hat is attributable to these three sectors in recent years,these so-called large firms account for somewhat over one-half and the giant-sized subset for almost one-fourth of the total(49,chap.1).FIG.11.1.Unitary form The question to be addressed is what problems does the U-form enterprise experience
17、 when it expands if the U-form(functional)basis for decomposing the enterprise remains in effect throughout?An answer that is both compelling and compact is not easy to provide.(For a more extensive discussion than is given here,see 49,chaps 2,3,7.)The two effects of enterprise expansion that are of
18、 most interest to us here are that expansion under the U-form structure(I)is subject to cumulative control loss,which has internal efficiency consequences,and(2)eventually alters the character of the strategic decision-making process.Finite spans of control naturally require that additional hierarch
19、ical levels be introduced as the U-form enterprise expands.Adding hierarchical levels can,if only for serial reproduction reasons,lead to an effective loss of control through incomplete or inaccurate transmittal of data moving up and instructions moving down the organizational hierarchy.Although var
20、ious decoupling devices may be devised to reduce these transmission needs,these are costly and subject to diminishing returns.Decoupling merely alleviates but does not overcome the need for intrafunctional and peak coordination.346 OLIVER E.WILLIAMSON Information flows rarely take the form of simple
21、 serial reproduction,however.Rather,data are summarized and interpreted as they move forward and instructions are operationalized as they move down.Both processes provide additional opportunities for control losses to develop.These can occur in quite unintentional ways.If,however,the functional unit
22、s of the firm view the hierarchical structure as affording opportunities to pursue local goals,deliberate distortions may be introduced into the hierarchical exchange process.Cumulative losses of these types stand eventually as an efficiency bar to the further expansion of the firm.Expansion also ev
23、entually overcomes the capacity of the office of the chief executive to provide strategic planning and maintain effective control.The usual means for augmenting this capacity has been to bring the heads of the functional divisions into the peak coordination process.The natural posture for these func
24、tional executives to take is one of advocacy in representing the interests of their respective operating units.This change in the composition of the strategic decision-making unit produces a shift away from preferences characteristic of the office of the chief executive,which tend to be enterprise-w
25、ide in scope,in favor of partisan interests more closely associated with the functional divisions.A persistent and collective pressure to provide more and better services is apt to develop;an expansionary bias in favor of staff expenditures easily obtains.A permissive attitude toward slack may also
26、result.The comparative static and dynamic-stochastic properties of a firm possessing these preferences have been presented elsewhere 43,46,49.For our purposes here it is sufficient to note that,ceteris paribus,progressive expansion of the U-form firm(1)results in greater control-loss experience,and(
27、2)the utility function of the firm is augmented to include the expense-preference inclinations of the functional divisions.Thus both internal efficiency and goal consequences are involved.2.Capital-market constraint Consider now whether,in an environment dominated by U-form enterprises,the capital m
28、arket can effectively restore selection on profit(that is,extinguish non-profit pursuits)in firms that would THE MULTI-DIVISION HYPOTHESIS 347 otherwise display discretionary tendencies.What controls does the capital market have access to?What limitations does it experience?In a general sense,the mo
29、st severe limitation the capital market encounters is that it is an external control device.This constitutes a more serious impediment than is often appreciated.The possibility of monitoring firm performance through the direct metering of funds is impaired by the internal financing which most firms
30、in the discretionary category have access to.The use of incentive devices of various sorts is impaired by the imprecision with which they apply and by the adaptive responses which they permit.The displacement threat which the capital market poses is subject to serious inference problems and experien
31、ces non-trivial displacement costs if the incumbent management is disposed to resist the takeover effort.(Again,for a more complete discussion,see 49,chaps 6,7.)Relevant in this connection is Shorey Petersons observation that Far from being an ordinary election,a proxy battle is a catastrophic event
32、 whose mere possibility is a threat,and one not remote when affairs are in conspicuous disarray.Indeed,even stockholder suits.may be provoked by evidence of serious self-dealing(33,p.21);emphasis added).But as recent history has made abundantly clear,atomic weapons with their catastrophic consequenc
33、es are ill suited to support military campaigns involving even half a million men.The principle is perfectly general:controls that have significant discreteness properties are appropriate only when an offense reaches egregious proportions.Otherwise,the remedy is too strong even to be invoked.Peterso
34、n concedes as much in his references to conspicuous disarray and serious self-dealing,and the evidence t!)nds to bear him out 22,46.Discreteness is significant,of course,because it implies that either the original or secondary costs associated with the use of a change agent are substantial.Where tak
35、eover is concerned,non-trivial costs ofboth types are involved.Not only are the original costs of securing displacement apt to be significant,but once displacement is accomplished the successful takeover agency needs to face the transition costs which a displacement involves.As more than one success
36、ful raider has come to appreciate,these can be neglected only at some peril.None of this is to suggest that the capital market in a U-form environment is wholly inefficacious in supplying incentives or enforcing lower-bound performance.Rather,the argument is that capital-market controls tend not to
37、be delicately conceived.348 OLIVER E.WILLIAMSON II.THE MULTI-DIVISION INNOVATION 1.Invention The historical development of the multi-division form has been traced by Alfred Chandler,Jr,in his superlative book titled Strategy and Structure.He reports that the M-form innovation was initially devised i
38、n the early 1920s,apparently quite independently,by Du Pont and General Motors;somewhat later,but still independently,it was adopted by Standard Oil of New Jersey and Sears 14.It has been widely imitated and rediscovered since.The immediate cause for the innovation to be introduced was the onset of
39、adversity.In Du Pont,the companys financial statement for the first half of 1921 provided the shock that finally precipitated a major reorganization.In those six months,as the postwar recession became increasingly severe,the company had lost money on every product except explosives(14,pp.126-7).At G
40、eneral Motors,an inventory crisis together with the collapse of the auto market in 1920 produced the change(14,pp.156-7).Partial reorganization at Jersey was induced by excessive inventories,falling profits,and a declining market share(14,p.230,but it was not until earnings fell to the lowest level
41、in 1927 of any year since 1912(from$117.7 million in 1926 to$40.4 million in 1927)that major organization changes were induced(14,p.256).Although profit pressures at Sears were less dramatic,they also contributed to the change(14,pp.321-3).The basic reason why the innovation became necessary,however
42、,is traceable to more fundamental reasons than transitory market conditions.Chandler summarizes the defects of the unitary form,and consequently the needs for the multi-division structure,in the following way 14,p.369):The inherent weakness in the centralized,functionally departmentalized operating
43、company.became critical only when the administrative load on the senior executives increased to such an extent that they were unable to handle their entrepreneurial responsibilities efficiently.This situation arose when the operations of the enterprise became too complex and the the problems of coor
44、dination,appraisal,and policy formulation too intricate for a small number of top officers to handle both long-run,entrepre-THE MULTI-DIVISION HYPOTHESIS 349 neurial,and short-run,operational administrative activities.To meet these new needs,the innovators built the multi-divisional structure with a
45、 general office whose executives would concentrate on entrepreneurial activities and with autonomous,fairly selfcontained operating divisions whose managers would handle operational ones.Illustrative of overloading conditions is the report of the troubles experienced by Du Pont following its diversi
46、fication moves but prior to its adoption of the multi-division form(14,p.111):Broad goals and policies had to be determined for and resources allocated to functional activities,not in one industry but in several.Appraisal of departments performing in diverse fields became exceedingly complex.Interde
47、partmental coordination grew comparably more troublesome.The manufacturing personnel and marketers tended to lose contact with each other and so failed to work out product improvements and modifications to meet changing demands and competitive developments.Each of the three major departments-Purchas
48、ing,Manufacturing,and Sales-made its own forecasts and set its own schedules.Thus,as the complexity of its various yet intervolved activities progressively increased,its ability to supply the requisite coordination became strained and even collapsed.Unable meaningfully to identify with or contribute
49、 to the realization of global goals,managers in each of the functional parts attended to what they perceived as operational subgoals instead(26,p.156).The experience of Ford in operating the giant River Rouge plant as a U-form organization in the 1920s affords a similar example(although,it should be
50、 noted,others have interpreted the problems of Ford mainly in personality rather than structural terms(32,chaps 8,11).The difficulties that the large,U-form enterprise became subject to(and which the M-form innovation was presumably designed to relieve)can be expressed in terms of indecomposability,
