Sunteți pe pagina 1din 14

The Structure of Small Bureaucracies Author(s): Peter M. Blau, Wolf V. Heydebrand, Robert E.

Stauffer Source: American Sociological Review, Vol. 31, No. 2 (Apr., 1966), pp. 179-191 Published by: American Sociological Association Stable URL: http://www.jstor.org/stable/2090903 Accessed: 09/02/2010 10:38
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=asa. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Sociological Association is collaborating with JSTOR to digitize, preserve and extend access to American Sociological Review.

http://www.jstor.org

THE STRUCTURE OF SMALL BUREAUCRACIES *


PETER

M.

BLAU, WOLF V. HEYDEBRAND AND ROBERT

E.

STAUFFER

The University of Chicago structureis based on data from 156 Americanpublic personnel An analysis of bureaucratic agencies.Whereasa professionalstaff was expectedto reduce the need for many managers, tends to lead to the centralizathe opposite is actually the case. Lack of professionalization tion of official authority in the hands of relatively few managers. The division of labor, which typically accompaniesgrowing size, promotes centralizationof authority only if the staff is not professional.This finding suggests that a centralizedauthority structureis well into simple routines but not for that of suited for the coordinationof tasks differentiated professionalspecialties.Structuralcomplexitiesdestroy the economicadvantage that operatof mechanisms communicalargescale otherwisehas, but bureaucratic ing on a comparatively staff,restorethis advantage. administrative tion, such as a sufficient

does Weber'sanalysisof bureau- are responsiblefor this joint occurrence. succession, analysisof managerial cratic structurecontinue to be re- Gouldner's gardedas the classic on the subject thoughbasedon a singlecase, vividlyshows of half a centuryafter it was written,despite how the exigencies the role of a new manthe many, often justified, criticisms that ager constrainhim to resort to formalized therebyhelpingto explainwhy have been directed against it? Does this procedures, tends to of replacement personnel merelyreflecta romanticregardfor one of recurrent the greatold menof socialtheory?We think be associatedwith formalizedmethods of is not. Weber'sgreat contribution that he operation.' Generallythe numerouscase studies of provided a frameworkfor a systematic that the theoryof formalorganization; fact that organizations have been carriedout in Weber's his analysishas certainlimitationsdoes not recent decadeshave complemented the achievement. analysisby investigating socialprocesses detractfrom this important particularlythe inshould within bureaucracies, structure A theoryof bureaucratic for Above all, it formal processesand their significance meet two basic requirements. At must be concernedwith the interdepend- the formalorganization. the same time, attributesof complex however,the focus of these studies on inence amongstructural and organizations not take these character- formalrelationsand practicesin one organistics of the structureas given and merely izationhas had the result that investigators examinethe decisionsor behaviorof indi- lost sight of the centralproblemof bureauvidualsin the context of complexorganiza- cratic theory, namely, what the interrelations. The focus of Weber'stheory is pre- tions betweenvarious structuralattributes are. This problem amongvarious of formal organizations ciselyon this interdependence different of characteristics the organizationsthem- can only be studiedby comparing and selves. In addition,a theoryof bureaucracy organizations not throughthe intensive between analysis of a single case. shouldaccountfor the connections theory,in contrast, At the coreof Weber's attributes by analyzing the organizational attributesof bureaucracy social processesthat have producedthese are the structural in It connections. is not enough,for example, and their relationships, the narrowsense to as to indicate,as Weberdid, that impersonal of "structural" referring the differenlines. alongvarious can in authority whichpersonnel readilybe tiationof socialpositions replacedand the use of formalizedproce- At the outset, Weberstressesthat responsiin "are distributed in durestendto occurtogether bureaucracies; bilities in a bureaucracy 2 we wouldalso like to know what processes a fixed way as officialduties," elaborating
W

Hy

I Alvin W. Gouldner,Patterns of IndustrialBu*We gratefullyacknowledgegrant GS-553 from the National Science Foundation. This is the first reaucracy,Glencoe:Free Press, 1954, pp. 45-101. 2From Max Weber: Essays in Sociology, New report of the ComparativeOrganizationResearch York: OxfordUniversity Press, 1946, p. 196. Programwhich it supports.

179

180

AMERICANSOCIOLOGICAL REVIEW
acteristics of the bureaucracies themselves, such as a status distribution in the organization, rather than merely attitudes or behaviors of the individuals in these bureaucracies. Moreover, the research design must involve the systematic comparison of a fairly large number of organizations, and not just a few cases assumed to be typical, in order to determine how variations in some characteristics affect variations in others. The important point is that such large-scale comparisons are required not only to test theoretical propositions once they have been formulated but also initially to formulate and refine the theory. When Weber portrayed a typical bureaucracy in bold strokes, on the basis of comparative analysis on a wide historical scale, he surely did not intend to suggest, as a simplistic interpretation sometimes assumes, that all the characteristics he outlined are highly correlated under all conditions. On the contrary, the complex interdependencies he traces clearly imply that the relationships between any two characteristics of bureaucracies often depend on and are modified by a third factor or even by a combination of several others. Our empirical data strongly confirm this crucial insight that the various aspects of bureaucraticstructure interact in their effects on each other. In other words, it is usually not possible to state simply what the relationship between two organizational attributes is, because it depends on one or more other attributes. We suspect that such higher-order interaction effects are a fundamental characteristic of social structures. In any case, given these complex interrelations in bureaucracies, the next step in refining the theory of formal organization is to specify the conditions under which different relationships hold. Such specification requires the quantitative analysis of concomitant variations in a fairly large number of cases. The organizations selected for the study are public personnel agencies, which are the executive agencies of the civil service commissions that administer the personnel policles of state and local governments. The sample consists of all members of the Public Personnel Association who returned a questionnaire about their agency in a survey ad-

later that this involves a systematic division of labor and often a high degree of specialization. Complementing this specialization is the requirement of "thorough and expert training,"3 which means that many positions in the organization are occupied by professionally or technically qualified specialists. The formal status hierarchy is a third fundamental characteristicof bureaucracy: "The principle of hierarchical office authority is found in all bureaucraticstructures." 4 Finally, the emphasis upon written communication and official documents in bureaucracies makes the employment of many clerks necessary, "a staff of subaltern officials and scribes of all sorts." 5 We propose to return to Weber's approach in this paper and to analyze the interrelations among these four structural attributes of bureaucracy and their implications for operations-the division of labor, professionalization, the hierarchy of authority, and the administrative staff of clerks. To be sure, Weber's analysis is not confined to these four but encompasses other formal characteristics of organizations that are not structural in the narrowsense of referring to aspects of status differentiations,6 such as the rules governing operations, the stable careers of officials,and the impersonalorientation that prevails in bureaucraticrelations. In this study, however,attention is restricted to the four organizational attributes indicated, and two others-size and an effect criterion. The investigation is based on data from about 150 public personnel agencies, which represent most of the larger organizations of this type in North America.
METHOD

The systematic study of bureaucratic structure necessitates a method of inquiry adapted to the purpose as well as an appropriate theoretical approach. The variables under investigation must be structural char8 Ibid., p.

198.

I The term "status"is used here in its broader sense as complementaryto role and synonymous with social position, not in the narrower sense distinctionsin social that restrictsit to hierarchical
position.

4 Ibid., p. 197. 5 Loc. cit.

SMALLBUREAUCRACIES in by ministered the Association 1958.7Returnswerereceivedfrom nearlyone half of the member agencies(252 of 528). Although it the sampleis not representative, includes agencies most of the largerpublicpersonnel in the United States and some in Canada,8 with the bias of self-selection working and againstsmalleragencies thoseleast idenof tifiedwith the meritprinciples civil service. All but seven of the 50 state agencies as arein the sample, aremostof the agencies cities,someof counof the largestAmerican agencies,such ties, and a few miscellaneous as that of the TVA. Public personnelagenciesare small buof Nearly two-fifths the 252 orreaucracies. had ganizationsrepresented a total staff of Thesewereexcluded fewerthanfivepersons. sincethe statusdistrifromthe investigation, bution measuresused have little validity if constructed sucha smallbase.This leaves on 156 cases for analysis,with a slightly lower total in most tables becauseof no answers. have Even after these tiny bureaucracies been eliminated,the median staff of the rest is only between16 and 17 persons,and havea staffof morethan a mere 17 agencies 100. Agenciesare divided into small ones with a staff of less than 20 (but at least 5) and largerones with a staff of 20 or more. This divisionclose to the medianseparates agenciesall of whose memberscan easily contactfromthose face-to-face havefrequent whose somewhatlarger size makes it unlikely that every memberknows all others well.9 are The six variablesunderconsideration

181

size, four attributes of internal structure,and a weighted measure of operating cost. The internal structure can be considered to have two major dimensions: specialization, which is a mechanism to deal with task complexity; and bureaucraticcoordination, which is a mechanism to deal with the organizational complexity introduced by specialization. Specialization can be subdivided into the division of labor in the organization and the degree of professionalizationof its staff. Two mechanisms of bureaucraticcoordinationare the hierarchy of managerial authority and the administrative apparatus. The followingare the operationalmeasures for these four structural attributes-two aspects of specialization and two kinds of coordinating mechanism: 1. Division of Labor: the numberof distinct occupationaltitles pertainingto the nonclericalstaff, not countingthose indicative of status differenceswithin a specialty Dichotoratherthan differentspecialties.10 mized betweenthree and four specialties.1' the 2.Professionalization: proportionof the as staff (excludingmanagers well operating as clerks) who are requiredto have, at

10Thus, "PersonnelTechnicianI" and "Personnel TechnicianII" were counted as one specialty. Since the absolute number of occupationaltitles is closely related to size, the question arises whether a relative index of the division of labor would not be preferableto an absolute one. But such a relative index-the ratio of occupationalpositions to size of staff-is just as much negatively associated with size as the absolute index is positively associatedwith it. 11In this case, the basic procedurefor determining the cutting points of the dichotomieswas modified for substantivereasons.The basic criterionfor dichotomizingwas to come as close to the median 7 These data were kindly made available to us as the initial categories on the IBM cards perfor statistical analysis by Messrs. Kenneth 0. Warnerand Keith Ocheltree,to whom we are in- mitted. Since the middle category was sometimes debted not only for doing so but also for helpful fairly large, and since elimination of the agencies with a staff of under 5 changed the proportions, advice. 8 There are only 14 Canadiancases, 4 provincial the numbers of cases in the two classes of the and 10 city agencies.Since tabulationsshowed that dichotomy are often not the same. Two exceptions these do not essentiallydiffer from the correspond- to the use of this criterion were made for coning agencies in the United States, they have not ceptualreasons.Division of labor was dichotomized been eliminatedfrom the analysisin order to maxi- between three and four to avoid calling agencies mize the number of cases. The U.S. Civil Service with only three different positions highly differand its Canadiancounterpartwere ex- entiated, as would be requiredif the median were Commission used, and professionalization was dichotomized cluded. 9Theodore Caplow specifies 20 membersas the above the median lest agencies with a single proupper limit of a primary group, "in which each fessionalon the operatingstaff be definedas highly memberinteractswith every other member";"Or- professionalized. These decisions about cutting ganizational Size," Administrative Science Quar- points were made before the substantiveanalysisof was carriedout. cross-tabulations terly, 1 (1957), p. 486.

182

AMERICANSOCIOLOGICAL REVIEW

differences reflectedin the denominator. are Whilethis is an adjustedmeasure operatof ing costs, it is not a reliablemeasureof efficiency,inasmuchas it does not take into accounteitherthe scopeof personnel services providedor the quality of performance. Althoughthe organizations underinvesti-. gation are not a representative sampleof a larger universe,statistical tests have been performedto furnish an external criterion A weighted measureof cost is employed to for decidingwhetheror not to place some investigate one of the implications of the confidence the complexrelationships in obvarious aspects of administrative structure served.YI use of these tests rests on the The for operations. The index of operating costs is the ratio of the salary budget of the per- assumptionthat a given pattern of differencesobserved wouldbe unlikelyto occurby sonnel agency itself to the total payroll for chancein a sampledrawnfroma hypothetithe entire civil -servicepersonnel under its cal universeof similarorganizations. jurisdiction-specifically, whether this ratio
exceeds one-half of one per cent or not. This AND MANAGERIAL PROFESSIONALIZATION measure adjusts cost not only to the magAUTHORITY nitude of the agency's operations but also to regional variations in standard of living. It Professional qualifications undoubtedly even takes into account, to some extent, that lessen the need for close supervision, other it is more difficult to administerhighly qual- thingsbeingequal.Wehadexpected therefore ified than unskilled personnel, since such that the ratio of managers operating to offi12Although agencies were divided on the basis

least, a college degree with a specified at major. Dichotomized 50 per cent.12 Hierarchy:the ratio of men in 3. Managerial posito managerial those in nonsupervisory tions among the nonclericalstaff, excludwas ing those for whomneitheralternative at indicated8 Dichotomized one to three.1 4. Administrative Apparatus:the proportion of clerksamongthe total staff.15 Dichotomizedat 60 per cent.

of whether less than 50, or 50 per cent or more, of the operatingstaff are requiredto have the specified professionalqualifications,the differencesare actually more extreme-since the distribution is bimodal. In more than four-fifths of the less professionalizedagenciesthere are no professionalrequirements for any position, and in more than three-quarters the professionalized of ones profesare sional qualifications requiredof the entire staff. The reasonfor excludingmanagerial personnelfrom this index is to assure that no relationship between professionalizationand management hierarchy can be due to lack of independenceof the two measures.Managers in professionalized agencies generallyare also requiredto meet professional qualifications. 18 Actually, the nonclerical personnel had been classified by respondents into four categoriesdeputy directors,heads of major divisions,journeymen, and apprentices. The index is the ratio of the first two to the last two. Positions that were not classifiedby respondentsmight be staff consultants or ambiguous cases. 14Sincethis excludesclerks,it corresponds conto siderably more than three subordinatesper manager-probably about seven. 15This is the only index that uses total staff in the denominator. The index provides a narrow operationaldefinitionof administrative staff; a wider one would include other positions responsiblefor maintainingthe organization,such as bookkeepers, but there are advantagesin restrictingit to clerical personnel,since they are the ones most directly with communication. concerned

cials would not be as high in organizations with a professional staff as in those where most of the staff is not requiredto meet professionalqualifications.The reasoning was that a high managerial ratio impliesa narrowspan of control,with few subordinates per manager, such a narrowspan and of controlis often assumedto be associated with close supervision. Contrary expectato tions, however,professionalized public personnel agenciesare more likely to have a high ratio of managers than other agencies under most conditions,though not under all. Unless one is willing to believe that professionalsare being more closely supervised than employees with less training, which seems improbable,this finding calls for a reappraisal the significance of of the managerial ratio. Instead of making
'sThe statistic developed by Leo A. Goodman has been used; "Modifications the Dorn-Stoufferof Tibbits Method for 'Testing the Significance of Comparisons Sociological in Data',"AmericanJournal of Sociology, 66 (1961), pp. 355-63. In a few cases, the criterion for the assumption of a normal distributionthat each cell contain at least 5 observations was not met; interpretation there should proceed with caution. We gratefully acknowledgeJohn Wiorkowski'sassistancein performing the statistical tests, and Leo Goodman's advice.

SMALLBUREAUCRACIES
HIERARCHY * TABLE PERCENTWITHCENTRALIZED 1. BY SIZEANDPROFESSIONALIZATION

183

in centralization the formal hierarchybut also on other conditions,such as the deleby gation of responsibilities superiorsto Professionsubordinates.No informationis available ** SmallAgencies LargeAgencies alization$** on these conditions,nor on closenessof suLow 38(61) 41(32) The ratiosinglesout for attention pervision. High 16(31) 43(28) an attributeof the formalstatus structure, Note: The differencein the left column is sig- not an aspect of the specifically, operations, nificant (at the 0.05 level). extent to whichauthoritypositionsare cen*The managerialhierarchy is centralizedwhen the ratio of non-clerical personnel in managerial tralizedratherthan being more widely dispositions to non-clerical personnel in non-super- persedin an organization.'7 visory positions, excluding those listed as neither, A centralized hierarchy of managerial is less than one to three; it is dispersedwhen the authorityis least likely to developin small ratio is one to three or more. as organizations, Table 1 professionalized ** Small agencies have a total staff of less than 20 (but five or more); large agencieshave a total shows. In small agencies with a professional staff, in other words, the probstaff of 20 or more. *** Professionalization low when the propor- ability of a high ratio of managers to is tion of the operatingstaff (excludingmanagersas subordinates greatest.Only 16 per cent is well as clerks) who are requiredto have, at least, a of theseagencies managehave a centralized college degree with a specifiedmajor, is less than ment (low ratio), whereasabout two-fifths 50 per cent; high when it is 50 per cent or more. small and of of the less-professionalized eitherkind of largeragenciesdo. Professionof a prioriassumptions about closenessof su- alizationand centralization authorityare let pervision, us start with the attributesof inverselyrelatedin smalleragenciesbut not bureaucratic structurethis ratio directlyre- in larger ones. Further analysis reveals, flects, and proceedwith the analysisbefore however,that these two factorsare also indrawinginferencesthat are consistentwith versely related in larger agenciesprovided that certainconditionsare met. the data. hierTable 2 indicatesthat a centralized The managerialratio indicates whether in is authority less prevalent rootedin formal archyof official the administrative authority in than in other organizations is in status in the hierarchy centralized the professional the handsof relativelyfew officials distributed three of the four comparisons, only exor To amonga largernumber. be sure,howcen- ceptionbeingthoselargeragenciesthat have Propersonnel. tralizedactualdecisionmakingis in the or- a low ratioof administrative the of reduces likelihood cennot ganization depends only on the degreeof fessionalization tralizationof authoritynot only in smaller ratio agencies,whateverthe administrative HIERARCHY (-24, -15), but also in largerones if the TABLE2. PER CENTWITHCENTRALIZED AND BY SIZE,PROFESSIONALIZATION, ADMINISTRAratio is high (-15), though administrative Tivz APPARATUS not if it is low (+18). The sharpest contrast SmallAgencies LargeAgencies is between the 19 smallerprofessionalized agencieswith a low administrative ratio, a Administrative Administrative mere11 per cent of whichhave a centralized * * Profession- Apparatus Apparatus hierarchy, the 15 largerprofessionalized and Low alization Low High High
Low High 35(31) 11(19) 40(30) 25(12) -15 29(17) 47(15) +18 53(15) 38(13) -15 "'The span of control is also indicative of the formal status hierarchy, and not of supervisory practice. The average span of control depends on the number of levels in the hierarchyas well as on the ratio of managers.No data on the number of levels are available. Though the ratio is not equivalentto the average span of control, the two are undoubtedly highly correlated,because variations in number of levels cannot be very large among organizationsfew of which have a staff of morethan 100.

Difference -24

in Note: The differencebetween differences cols. 1 and 3 is significant (at the 0.05 level); that between cols. 2 and 4 is not.
* The administrative apparatus ratio is low when

the proportionof clerksamong the total staff is less than 60 per cent; high when it is 60 per cent or
more.

184

AMERICANSOCIOLOGICAL REVIEW

the ratioof managers, with few subordinates for each, the greaterare the opportunities for frequentdiscussionsin which problems expressed, can be explored,dissatisfactions and conflictsreconciled. A high managerial ratio, whichimpliesa hoc interpretations. promotesextensive dispersedmanagement, trainingmay be assumedto verticalcommunication the hierarchyof Professional in communiin makea manmoreself-directing his work. authority.It facilitatesdownward that cation, which makes it easier for superiors It was on the basis of this assumption and as we had expected-incorrectly, it turned to directthe workof subordinates check staff up on them. But it also facilitatesupward out-that agencieswith a professional Identi- communication, which makes it easier for wouldhave a low ratio of managers. to however, subordinates convey informationto sustandards, fication withprofessional detends to makea man workingin an organ- periorsand to influenceadministrative ization not only less dependenton direct cisions. Our original expectationhad exsupervisionbut also more aware of the clusively focused on the significanceof a broaderimplicationsof his job and more high managerialratio for downwardcomwhereas datasuggestthat its the in interested seeingto it that agencypolicies munication, is and operationsdo not violate professional significancefor upward communication upward principles.Such a man is likely to detect the morecrucialhere.To encourage a by that escapethe noticeof one with- communication appointing highproporproblems important and qualifications to want tion of managersis particularly out professional with a professional staff, modified for organizations procedures to have administrative The to remedytheseproblems. professional's becauseonly by doingso can they take full of professionals with helpingshapeagency advantage the contributions concern informed and procedures policiesis at the root of the are capableof making. of Professionalization and centralization and conflictbetweenprofessionals recurrent a but administrators, it is simultaneously re- authorityappearto be alternativemethods Since profesresponsibilities. for sourceavailableto the organization im- for organizing enhancea man'sability sionalqualifications provingoperations. of Management cannot give professionals to see the implication his workand place can free rein,since thereis often a genuinecon- it into a wider context, professionals in and professional administrative contributeto coordination an organizaflict between is Yet seriously tion; the taskof management to drawupon considerations. if management and frustratesprofessionalsin the exercise of thesecontributions fit theminto the adThis requires sufa framework. it their responsibilities, not only courts the ministrative of to danger of dissatisfactionsand defections ficientnumber managers workin close with the professional staff. If but fromthe organization also fails to take collaboration the staff lacks professional training,on the advantageof an importantresourceat its disposal,which includes the professional's other hand, it can make only limited conand interestin perfectingoperationsas well as tributionsto coordination, the task of to is, to management consequently, effect cohis expert knowledge.For management largelyon its own ratherthan to into ordination channelthe initiative of professionals with operatingofficialsfor this administrativeimprovements instead of collaborate contactand close purpose.A hierarchyin which authorityis it requires frequent stifling in betweenmanagers profes- concentrated the hands of relativelyfew and collaboration in functions such servesdistinctive sionals. Though conflicts between profes- managers a situation,becauseit makesit possibleto concernsare insional and administrative evitable,the best chance for advantageous achieve coordinationthrough centralized groupand by is compromise probablyprovidedby exten- planning a smallheadquarters to re- then issuingpertinentdirectives the staff. betweenthe officials sive communication and sponsiblefor professional those respon- The limitations on discretion centralized imposeuponthe staff and The decisions. higher planning direction sible for administrative

ratio,53 with a high administrative agencies to percentof whichdo.The twoquestions be answeredare, first, why the negative relationshipprevailsin most situationsand, second,what accountsfor the deviantcase. The answerssuggestedare admittedlypost

SMALLBUREAUCRACIES are undoubtedly more objectionable proto fessionals than to men whoselack of expertnesswouldmakeit hardfor themto exercise much discretion.To be sure, a centralized structure makesit moredifficult the relafor tively few managers keep in close touch to with operatingofficialsand supervisethem closely. But modernadministration dehas vised substitutemethods for obtaininginformation operations checking the on and on work of subordinatesthat do not require frequent direct contact, such as detailed statisticalrecordsof performance.'8 In the absenceof a professionally trained staff, accordingto these considerations, a centralizedhierarchyof authorityhas importantadvantages an organization for and comparatively disadvantages. major few Its advantageis that it can meet the need for coordinationthrough centralizedplanning. In a relativelylarge organization, however, this requires adequate an administrative apparatusof clerks to maintainthe channels of communication are essentialfor cothat froma centralheadquarters. ordination This interpretation bringsus back to the findings in Table 2, becauseit suggestswhy a centralizedhierarchy morelikely to develop is in agenciesthat arenot professionalized than in professionalized if they are small or ones if they are largerand have a high administrativeratio,and why this is not the case for larger agencies with a low administrative ratio. Professionalization well as cenas in tralization, turn, dependon the division of labor.
STRUCTURAL DIFFERENTIATION

185

vanceddivisionof labor with four or more occupationalspecialties is found in only 14 per cent of the 92 smallpublicpersonnel agenciesbut in 66 per cent of the 62 larger ones. A minimum size is virtuallyrequired for the development severaldistinctfuncof In tionalpositionsin an organization."' contrast to this stronginfluence the division on of labor, size exerts little influenceon the threeotherbureaucratic attributes underinratio vestigation. Thus,a highadministrative is as likely to be found in largeragencies as in small ones.20There is no significant difference between largerand smallagencies, moreover, eitherin the likelihoodthat their staff is professionalized in the likelihood or that their authoritystructure centralized. is Althoughsize has no direct effect on these factors,it has distinct indirecteffects that modify their interrelationsand condition their significance operations. for which is typiStructuraldifferentiation, cally a consequenceof expandingsize, in of turn affects other characteristics the organization.It has implicationsfor professionalization and for centralization: the pattern revealedby the resultingrelationthat ships supportsthe previousconclusion professionalization and centralizationare the alternativemodes of organization, existenceof one beingsomewhat incompatible with that of the other.The divisionof labor or promoteseither professionalization centralizationbut not both, and whether it
19 In the extreme case, such a relationshipis a mathematicalnecessity-agencies of less than four personscould not possiblyhave four or more positions-but none of the agencieshad a staff of less than five. 20Whereas other studies of organizationsfound that the administrative ratio declines with size, this is the case only after a certain size has been reached.This is probablythe reason for the difference in findings-since our data are primarilybased on fairly small organizations. SeymourMelman, See "The Rise of Administrative Overhead in the Manufacturing Industries of the United States, 1899-1947," Oxford Economic Papers, 3 (1951), pp. 89-90; ReinhardBendix, Work and Authority in Industry, New York: Wiley, 1956, pp. 221-222; TheodoreR. Andersonand SeymourWarkov,"Organizational Size and Functional Complexity," American Sociological Review, 26 (1961), pp. 23-28; and Wolf V. Heydebrand,"Bureaucracy Hospiin tals; An Analysis of Complexityand Coordination in Formal Organizations," unpublishedPh.D. dissertation, Department of Sociology, University of Chicago,1965.

Expansion sizehas a pronounced in impact on the structural differentiation functional of specialtieswithin the organization. adAn
18 One might even speculate whether the very fact that a low ratio of managersfacilitates close supervisiondoes not make it particularlyinappropriate for a nonprofessional staff. The experienced manager in charge of a nonprofessionalstaff is more likely to be tempted to supervisetoo closely if conditionspermit than is the managerin charge of expert professionals. Since a low ratio facilitates close supervision(indicated by researchto be detrimentalto performance),and since the inclination to resortto close supervisionis greatestif the staff is not highly skilled, a low ratio of managersmay be most dysfunctionalfor operationswith an unskilledstaff.

186
TABLE PERCaiTw 3. wTv

AMERICANSOCIOLOGICAL REVIEW
CErT ERmD HxEAPaCn BY PROFESSIONALIIZATIOV ANDDIvISIoN

OF LABOR

the the stafflacksprofessional qualifications, proportion of centralized hierarchies increases with growing task differentiation

Professionalization Divisionof Labor Rudimentary Advanced Low 33(64) 50(28) High 31(36) 26(23)

in smalleragencies (? 14) as well as in larger ones (+32). When the staff meets professionalrequirements, contrast, the in proportionof centralized hierarchiesdecreases with growing task differentiation in

Note: The differencein the left column is not significantat the 0.05 level. *The division of labor is advanced when the non-clericalstaff representsfour or more occupational specialties; rudimentarywhen it represents less than four.

promotes one dependson the absenceof the the other.Thus, the chancesthat the hierarchy of authoritybecomescentralized increasewith advances the divisionof labor in only if the staff is not professional, not and at all if it is professional, Table 3 shows. as Aboutone-third the agencieswith a rudiof mentarydivisionof laborhave a centralized structure,whatever the degree of profesand sionalization, even slightly fewerof the more differentiated agencies that are professionalized, comparedto one-half of the more differentiated agencies that are not professionalized. In brief, task differentiation seems to furtherthe development a of centralizedhierarchy of authority unless the organization professionalized. is The conclusionthat the influenceof the divisionof labor on the management hieron archydepends professionalization howis, in ever,suspect.The relationship Table 3 is not significant, and size has not been controlled,although is knownto havea strong it impact on the division of labor and some on But bearing the two othervariables. controllingfor size does not destroythe pattern it observed;on the contrary, accentuates it. In bothsmaller largeragencies, Table and as 4 indicates, intensification the division an of of laborraisesthelikelihood the emergence of of a centralized if hierarchy the staff is not professionalized, it actuallylessens this but likelihood it is professionalized; if thesedifferencesare statistically significant.When
when percent21The pattern is complementary ages for the data in Table 3 are computed with as professionalization the dependentvariable. Difin ferentiationtends to promoteprofessionalization the absence of centralizationbut not when it is present.

bothsmaller(-19) andlarger(- 27) agencies. Holdingsize constantmagnifiesthe interaction effectof differentiation profesand sionalization centralization. confirms on This the conclusion that structural differentiation promotes emergence a centralized the of hierarchyonly in the absenceof professionalization; in its presence,it does not and may even have the opposite result. The divisionof labor in an organization can take two entirely differentforms. On the one hand,it may involvethe subdivision of the overalltask of the organization into specialized responsibilities that permit,and indeedrequire,greaterutilizationof expert specialists.This development illustrated is by the difference betweena hospitalmedical staff consistingof generalpractitioners and one composedprimarilyof specialists. On the otherhand,the divisionof labormay entail the fragmentation responsibilities of into with routineduties that simpleassignments requireminimal skills. The assembly line factoryis an extremeexampleof this tendency. Since many professionals essential are in the one case while few are neededin the other, the extent of professionalization a in publicpersonnel agencyindicateswhichone of these two forms the division of labor
TABLk4. PER CENTWITS CENTRALIZED HIERARCHY BYSIZE,DIvIsIoN OSLABOR, AND PROFESSIONALIZATION

SmallAgencies Division of Labor

LargeAgencies

ProfessionalizationProfessionalization Low Low High High 19(26) 0(5) -19 18(11) 50(20) +32 60(10) 33(18) -27

Rudimentary 36(53) Advanced 50(8) Difference +14

Note: The differencebetween the pooled differences in columns 1 and 3 and those in columns 2 and 4 is significantat the 0.05 level. (One-half observation, or ten per cent, was substitutedfor the
zero cell, because variance cannot be estimated by the method used, for cells with zero observations.)

SMALLBUREAUCRACIES
has taken. Specialized differentiationof tasks (an advanced division of labor in combination with professionalization) and routinized differentiationof tasks (an advanced division of labor in the absence of professionalization) pose different administrative problems. Routinized differentiation, which minimizes the need for professional experts, maximizes the need for managerial coordination. Centralizedplanning and direction are effective means for coordinatingfragmentedduties performed by a relatively untrained staff. Routinized differentiation accordingly enhances the likelihood of the development of a centralized authority structure.22 Specialized differentiation does not pose the same problem of coordination for management as routinized differentiationdoes, because tasks are not as fragmented, because professionals are qualified to assume wider responsibilities, and because managementgains advantages from eliciting their contributions to coordinationinstead of imposing directives from a central headquarters on them. A centralized management has great disadvantages in a professionalized organization, as has been noted. Hence, specialized differentiation does not enhance the likelihood of the development of a centralized hierarchy. In sum, structural differentiationin public personnel agencies sometimes is accompanied by greater professionalizationand sometimes by greater centralization of authority, and whether it leads to the elaboration of one of these depends in part on the absence of the other of these two alternative modes of organizing responsibilities. An advancing division of labor that is associated with professionalization raises entirely differentadminis22 The conclusion that routinized differentiation creates a need for centralization,and the earlier conclusionthat centralization larger agenciesdein pends on an adequateadministrativestaff, together imply that routinizeddifferentiation should lead to the expansionof the administrativeapparatus.The data give some support to this inference.The proportion of agencies with a high administrative ratio increaseswith growing differentiationneither in small agenciesnor in larger professionalized ones but only in larger agencies lacking professionalization, from 18 per cent of 11 cases to 60 per cent of 20. This differencebarely fails to reach the conventional level of statistical significance,being significant at the 0.07 level.

187

trative problems from one that is not. The former makes adequate upward communication especially important, without which management is deprived of some of the contributions the professional staff can make. To encourage upward communication requires a high ratio of managers dispersed throughout the organization.The subdivision of labor among a staff lacking professional qualifications, in contrast, makes adequate downward communication particularly important, without which the coordination of diverse simple routines cannot be accomplished. A centralized hierarchy of authority facilitates such coordination, provided that it is complementedin organizationsbeyond a minimum size by a sufficient administrative staff of clerks to maintain the essential lines of communication.These considerationssuggest that, in an organization, a high ratio of managers is of special importance for upward communication and a high ratio of clerks for downwardcommunication.
IMPLICATIONS FOR OPERATING COSTS

How do various conditions in the bureaucratic structure affect operating costs? To be sure, it would be of great interest to examine the significanceof differencesin the administrative structure for other aspects of operations as well as cost, but the data necessary for this purpose are not available. There is no denying the importanceof budgetary considerations in government agencies and most other formal organizations; the study of operating costs is consequently a good starting point for clarifying the implications of bureaucratic attributes for operations. The weighted measure of operating costs, to repeat, is whether or not the salary budget of the public personnel agency exceeds one-half of one per cent of the total salary budget for the civil servants under its jurisdiction. Structural differentiation, on the whole, lowers operating costs. Whereas exactly onehalf of the 98 agencies with a rudimentary division of labor operate at relatively high costs, 35 per cent of the 54 with an advanced division of labor do. Since larger agencies also operate on the average at lower costs than smaller ones, and since size and the division of labor are strongly related, the question arises whether task differentiation or operating on a large scale actually pro-

188

AMERICANSOCIOLOGICAL REVIEW
ADi DIVISION LABOt OF

hoodof high costs (from60 to 28 per cent), and professionalization larger organizain tions has, if anything,a detrimental effect Division economy,in sharp contrastto of Labor SmallAgencies LargeAgencies on operating its beneficialeffect in small organizations. 33(21) Rudimentary 55(77) These data suggest that professionalization, Advanced 23(13) 39(41) as well as differentiation, two contradichas Note: The differencein the left columns is sigtory implications operations. for nificantat the 0.05 level. Differentiation professionalization and in* Agencies in which the salary budget of the agency itself is 0.5 per cent or more of the total fluence operationsdirectly, and they also payroll for the entire civil service personnel the produce changesin the organizational strucagency administers. ture that have indirect repercussions for operations. The direct effectsare most eviduces these cost reductions. The answeris dent in smaller organizations, and the inthat both lead to economies, Table S re- directones,whichhaveopposite as implications veals, but their effects are not cumulative. for operations, largerorganizations. in Task Small undifferentiated agencies are most differentiation the manifestpurposeof has likely to operateat high costs. Eitheran in- raising efficiency, it achieves purpose and this crease in differentiation an increasein in small agencies.At the same time, howor size tends to reduce operatingcosts, but ever, differentiation increases the internal the occurrence both doesnot reducethem complexityof the organizational of structure. which While the immediate purpose of profesfurther. Althoughtask differentiation, usually accompaniesgrowth, effects econ- sionalization to improve is performance qualomies, it simultaneously destroys the eco- ity, the findingsindicatethat it also raises nomic advantagethat operatingon a large efficiency small agencies. in The reasonmay scale otherwisehas (the right-handvalue be that the abilityof professionals fit their to is the lower one in the first but not in the owntasksintoa widerframework contributes secondrow of the table). to the overallcoordination organizations in for sufficiently The significance professionalization of small to permitregularface-tooperating costs closely parallelsthat of the face contacts among the entire staff, but divisionof labor.Table 6 showsthat opera- much less so in larger organizations with tions are most likely to be costly in small severaldepartments. Separatedepartmental Professional- groupsof professional with few professionals. agencies specialists increase the of izationgreatlylessensthe likelihood high structural of complexity an organization, just costs in small organizations (from 60 to 31 as does the divisionof labor.Internalcomper cent), thoughonly there,just as is case plexity gives rise to problems,which are for differentiation. findingthat profes- reflectedin operating The costs, as can be seen sionals,despitethe highersalariesthey com- most clearly when complexityis viewedas in mand,lowerthe cost of operations small a conditionthat modifiesthe basic influence agencies impliesthat theymakecontributions of size on operating costs. to administrativeefficiency-apart from Simple agencies exhibit an economy of thosetheirexpertskillsmaketo performance scale,whereas onesdo not. Whether complex quality,whichare not reflectedin the cost the divisionof laboror professionalization is on measure. Operating a largerscalewithout taken as the indicationof structuralcomprofessionals, however, lessensthe likeli- plexity, largerorganizations also tend to operate at lower costs than smaller ones if their TABLE 6. PxR CENT WITH HIGH COSTS BY SIZE AND structureis simple, but not if it is comPROPEssIONALIZATION plex (see Tables 5 and 6). Internal complexity nullifiesthe economic advantage opProfessionSmallAgencies alization LargeAgencies erations on a large scale have in its absence, becauseit gives rise to problemsof 28(32) Low 60(61) in and coordination larger communication 43(28) High 31(29) organizations. While this is self-evidentin Note: The differencein the left column is sigthe case of the divisionof labor,it requires nificantat the 0.05 level.
TABLE 5. PER CENT WITH HIGH COSTS* BY SIZE

SMALL BUREAUCRACIES

189

OF BY AND ADMINISTRATIVE HIGH COSTS SIZE,DIVISION LABOR, APPARATUS TABLz7. PERCENTWITH


Division of Labor and Administrative Apparatus Rudimentary: Low Administrative Ratio High Administrative Ratio Advanced: Low Administrative Ratio High Administrative Ratio Small Agencies 51(41) 58(36) 12(8) 40(5) Large Agencies 36(14) 29(7) 50(20) 25(20) Difference -15 -29 +38 -15

Note: The difference between the differences in rows 1 and 3 is significant at the 0.05 level; that between rows 2 and 4 is not (0.63).

explanation in the case of professionalswhose coordinating ability has been noted. Professionals who are "locals" and identify themselves with the organization in which they work have been distinguished from "cosmopolitans,"who are primarily oriented to the wider group of professional colleagues anywhere.28It seems reasonable to assume that locals predominate in smaller organizations and cosmopolitans in larger ones. The broader context into which professionals tend to fit their own tasks is the. work of the organization, for locals, but the work of fellow specialists outside, for cosmopolitans. The very identification of professionals with their responsibilities that leads the locals in an organization to converge in their orientations leads the cosmopolitans to diverge. Professionals in larger organizations, often working in diverse departments and oriented to differentprofessions outside, consequently intensify problems of communication. The interpretation suggested is that the problems of communicationproduced by professionalization as well as by differentiationin larger
23 Alvin W. Gouldner, "Cosmopolitans and Locals," Administrative Science Quarterly, 2 (1957/58), pp. 281-306, 444-480.

agencies account for the adverse effects of these structural complexities on operations. This interpretation implies that complexity no longer has adverse effects once mechanisms to deal with problems of communication have been developed. The data support this inference. Organizational complexities destroy the economy of scale that is otherwise observable, but appropriate bureaucratic mechanisms for coping with the problems posed by these complexities restore the economy of scale. Table 7 reveals that, among simple organizations with a rudimentarydivision of labor, a larger scale of operations tends to reduce costs, whether the administrative ratio is low (-15) or high (-29). Among complex organizations with an advanced division of labor, however, a larger size tends to raise costs if the administrative ratio is low (+38) and reduce them only if it is high (- 15). The same pattern appears when professionalization is substituted for the division of labor, as shown in Table 8. In the absence of professionalization, an increase in size lessens the likelihood of high costs, whatever the administrative ratio (-31, -30), but in professionalized agencies, an increase in size raises the likelihood of high

TABLE

8.

AND APPARATUS PER CENT WITH HIGH COSTS BY SIZE, PROFESSIONALIZATION, ADMINISTRATIVE

Professionalization Apparatus and Administrative Low: Ratio Low Administrative Ratio High Administrative High: LowAdministrativeRatio Ratio High Administrative

SmallAgencies 55(31) 63(30) 28(18) 36(11)

LargeAgencies 24(17) 33(15) 60(15) 23(13)

Difference -31 -30 +32 -13

Note: The differencebetween the differencesin rows 1 and 3 is significantat the 0.05 level; that between rows 2 and 4 is not.

190
Divisionof Labor andManagerial Hierarchy Rudimentary: Dispersed Hierarchy Centralized Hierarchy Advanced: Dispersed Hierarchy Centralized Hierarchy

AMERICANSOCIOLOGICAL REVIEW

ANDMANAGERIAL HURARCHY TO=rac PER CENT WITH HiGH CosTs BY SizE, DIvISION OF LABOR, 9.

SmallAgencies 50(54) 65(50) 11(9) 50(4)

LargeAgencies 23(13) 50(8) 39(23) 39(18)

Difference 27 -15 +28 -11

Note: The differencebetween the differencesin rows 1 and 3 is significantat the 0.05 level; that between rows 2 and 4 is not.

ratio cost$ (+32) unless the administrative is high ( -13). An adequateadministrative apparatusto Maintainchannels of communication can meet the problems created by structural in complexity organizations beyonda minimumsize; therebyit reinstatesundercomplex conditions economythat accruesto the large-scaleoperationsunder simple conditions without it. A centralizedauthority structure serves equivalent functions, aland thoughonly for task differentiation not for professionalization. the division of If labor is rudimentary, Table 9 indicates, as operating costs tend to declinewith expandof ing size regardless the management hierarchy (-27, -15) but if the division of
labor is advanced, costs tend to rise with

lower operatingcosts, as manifest in the data from small agencies, but they simultaneously increase the structural complexity of the organization. Operating on a larger scale also lowers operating costs, as revealed by the data from simple agencies, though not by those from complex ones (because of the disturbing influence of complexity). Structural complexity raises problems of communication in larger organizations which, if unresolved, impede effective operations, as implied by the findings that complexity eliminates the economic advantage of larger agencies but administrative mechanisms that meet problems of communicationre-establish this advantage.
CONCLUSIONS

expanding size (+28) unless the hierarchy is centralized (-11). Centralizationof The interrelationsamong four bureaucratic formalauthorityfacilitatesthe coordination attributes in Americanpublic personnelagenof diverse tasks-a major problemin dif- cies have been analyzed-task differentiation, ferentiated largerorganizations. pattern professionalization, the management hierThe of findingsin Table 9 corresponds closely archy, and the administrative apparatus. It withthosein Tables7 and8. Sincecentraliza- is noteworthy that all six zero-order relation and administrative apparatusare not tionships between any two of these four significantly related,and neitherare division organizational attributes are insignificantly of labor and professionalization, three small, not one making a differenceof as much the are but sets of findings not redundant actual as 12 per cent. Even when size is controlled replications. combined The influence pro- only one of the six reveals a significant difof fessionalization centralization and does not, ference (that betweenprofessionalizationand however,reveal the same pattern. This is centralization). It is primarily in the higherhardlysurprising the light of the earlier order interactions that distinct relationships in indicationsthat a centralizedhierarchyis become apparent. This creates methodologinot a suitable coordinating mechanismfor cal difficulties, since many more than the a professional staff. 150 cases of organizations here available In short, proceduresinstituted to meet would be needed to explore adequately the some problemsoften have repercussions intricate interrelations. But it is also of in the organizational structurethat createnew substantive significance, for these higherproblems.These conflictinginfluenceshad order interactions reflect and provide emto be inferred from the analysis.Thus, both pirical validation for the theoretical conthe divisionof laborand professionalization ception of social structure, which implies a

SMALL BUREAUCRACIES

191

complexinterdependence between elements A systematic analysis of bureaucratic here,seeks rather than correlationsbetween pairs of structure, the one attempted like attributesunaffectedby other conditions.24 to discover the consistent interrelations Professionalization plays a dual role in among organizational attributes.Consistent the bureaucratic structure,being in some patternscannot be found in all the data; respects the counterpart the divisionof those that can are singledout for attention. of labor, and in others that of the authority While such selectivity is inevitablein exhierarchy. Just as doesan advanced division ploratoryresearch, creates the dangerof it of labor,professionalization lowerscosts,en- conveying impression greater of functional an gendersstructural complexities produce integrationin the structure than actually that communication problems,and requiresthe exists. It should be mentionedas a caveat, assistanceof an administrative staff. On the therefore, that functional relationships, other hand, it seems to be an alternative thoughoften implicitin the analysis,by no to a centralized authority structure. Whereas means prevail throughoutthe bureaucratic the analysisof operatingcosts impliesthat structure. elementsthat Thereare numerous professionalization increases problems of revealno positive feedbackfrom their concoordination communication, inverse sequencesto make them functionalfor opand its relationshipto centralization suggests that To it helps meet problemsof coordination, but erations. cite only one example:an increase in size promotestask differentiation only if complemented a dispersedmanby of agement. Although these findings appear but not the expansion the administrative staff. This is a dysfunctionalconsequence, contradictory, canreadilybe reconciled. they lowers operating The inverse relationshipbetween profes- since task differentiation and sionalization centralization means,after costs only in small agenciesbut raisesthem by all, nothingelse thanthat a professional staff in larger ones unless supplemented an administrative staff. adequate is usually accompanied a large and disby suggestedby the The generalconclusion persedmanagerial this component; does not in the least conflictwith the interpretation analysis is that the complexinterrelations that professionalization in intensifiesproblems andhigher-order interactions observed the of communication. be sure,professionals organizational To structureare more likely to also make some contributions coordina- be functionallyadapted by feedback than to tion, particularly detectingproblems by and the separate attributes themselves. Feedproposing knowledgable solutionsfor them. backprocesses not seem to produce so much For these contributionsto be realized in the elimination dysfunctional elementsas of effectivecoordination, however,an adequate a greaterfunctionalinterdependence among staffof managers essentialto workin close themin the bureaucratic is structure. refined A contact with professionals,and to imple- conceptionof functionalinterdependence is ment theirproposals well as solicit them. implicit here, which would neither simply as The effectiveness a professional of staff, in meanthat the elementsin a socialstructure sum, dependson its being complemented by nor assume that each one an adequatemanagerial and in are interrelated component but of themservesimportant functions, which adlargerorganizations by an adequate also to wouldrefer specifically the fact that the These are required ministrative component. relationshipsamong and complex, higher-order to meet the problemsof coordination in have been adapted elements the structure communication professionalizationraises, The notwithstanding ability of professionals by feedbackto minimizedysfunctions. the is interdependence functional,althoughthe to help solve these problems. specific factors may not be. External cir24 Hubert M. Blalock, Jr., stressesthe importance cumstances and vestedpowersoften impose of taking explicitly into account such higher-order conditionson organizations that are dysinteractionsin sociologicaltheory; "TheoryBuild- functional operations; for feedback processes ing and the Statistical Concept of Interaction," can at best minimizedysfunctionswithin American Sociological Review, 30 (1960), pp. this frameworkbeyond their control. 374-380.

S-ar putea să vă placă și