Poiver adheres to those who can cope ivith the critical problems of theorganization. As such, power is not a dirty secret, but the secret of success.And that's the path power follows, until it becomes institutionalized—which makes administration the most precarious of occupations.

Who Qets Power-MdMowZkey Mold OH to

A Stmtegic-ContiHgemtiModel of Power

Gerald R. Salancik

Jeffrey Pfeffer

^ wordis held by many people to be a dirty

word or, as Warren Bennis has said, "It is theorganization's last dirty secret."

This article will argue that tradi-tional "polidcal" power, far from being adirty business, is, in its most naked form,one of the few mechanisms available foraligning an organization with its own reality.However, institutionalized forms of power—what we prefer to call the cleaner forms ofpower: authority, legitimization, centralizedcontrol, regulations, and the more modem"management information systems"—tendto buffer the organization from reality and

obscure the demands of its environment.Most great states and institutions declined,not because they played politics, but becausethey failed to accommodate to the po-lidcal realities they faced. Political processes,rather than being mechanisms for unfair andunjust allocations and appointments, tend to-ward the realistic resolution of conflictsamong interests. And power, while it eludesdefinition, is easy enough to recognize by itsconsequences—the ability of those who pos-sess power to bring about the outcomes theydesire.

The model of power we advance is

Organizational Dynamics, W'mter 1911. © 1911, AMACOM, a dhiston ofAmerican Managemem Associations. All rights reserved.

an elaboration of what has been called stra-tegic-contingency theory, a view that seespower as something that accrues to organiza-tional subunits (individuals, departments)that cope with critical organizational prob-lems. Power is used by subunits, indeed,used by all who have it, to enhance their ownsurvival through control of scarce criticalresources, through the placement of allies inkey positions, and through the definition oforganizational problems and policies. Be-cause of the processes by which power de-velops and is used, organizations becomeboth more aligned and more misaligned withtheir environments. This contradiction is themost interesting aspect of organizationalpower, and one that makes administrationone of the most precarious of occupations.

WHAT IS ORGANIZATIONAL POWER?

You can walk into most organizations andask without fear of being misunderstood,"Which are the powerful groups or peoplein this organization?" Although many orga-nizational informants may be umvUUng totell you, it is unlikely they will be unable totell you. Most people do not require explicitdefinitions to know what power is.

Power is simply the ability to getthings done the way one wants them to bedone. For a manager who wants an increasedbudget to launch a project that he thinks isimportant, his power is measured by hisability to get that budget. For an executivevice-president who wants to be chairman,his power is evidenced by his advancementtoward his goal.

People in organizations not onlyknow what you are talking about when youask who is influential but they are likely toagree with one another to an amazing extent.

4 Recently, we had a chance to observe this in

a regional office of an insurance company.The office had 21 department managers; weasked ten of these managers to rank all 21 ac-cording to the influence each one had in theorganization. Despite the fact that ranking21 things is a difficult task, the managers satdown and began arranging the names oftheir colleagues and themselves in a column.Only one person bothered to ask, "Whatdo you mean by influence?" When told"power," he responded, "Oh," and went on.We compared the rankings of all ten man-agers and found virtually no disagreementamong them in the managers ranked amongthe top five or the bottom five. Differencesin the rankings came from department headsclaiming more influence for themselves thantheir colleagues attributed to them.

Such agreement on those who haveinfluence, and those who do not, was notunique to this insurance company. So far wehave studied over 20 very different organiza-tions—^universities, research firms, factories,banks, retailers, to name a few. In each onewe found individuals able to rate themselvesand their peers on a scale of influence orpower. We have done this both for specificdecisions and for general impact on organi-zational policies. Their agreement was un-usually high, which suggests that distribu-tions of influence exist well enough in every-one's mind to be referred to with ease—andwe assume with accuracy.

WHERE DOES ORGANIZATIONAL

POWER COME FROM?

Earlier we stated that power helps organiza-tions become aligned with their realities.This hopeful prospect follows from whatwe have dubbed the strategic-contingenciestheory of organizational power. Briefly,those subunits most able to cope with the

organization's critical problems and uncer-tainties acquire power. In its simplest form,the strategic-contingencies theory impliesthat when an organizadon faces a number oflawsuits that threaten its existence, the legaldepartment will gain power and influenceover organizational decisions. Somehowother organizational interest groups willrecognize its critical importance and conferupon it a status and power never before en-joyed. This influence may extend beyondhandling legal matters and into decisionsabout product design, advertising produc-tion, and so on. Such extensions undoubtedlywould be accompanied by appropriate, oracceptable, verbal justifications. In time, thehead of the legal department may becomethe head of the corporation, just as in timespast the vice-president for marketing hadbecome the president when market shareswere a worrisome problem and, before him,the chief engineer, who had made the pro-duction Une run as smooth as silk.

Stated in this way, the strategic-contingencies theory of power paints an ap-pealing picture of power. To the extent thatpower is determined by the critical uncer-tainties and problems facing the organiza-tion and, in turn, influences decisions in theorganization, the organization is aligned withthe realities it faces. In short, power facili-tates the organization's adaptation to its en-vironment—or its problems.

We can cite many illustrations ofhow influence derives from a subunits's abil-ity to deal with critical contingencies. Mi-chael Crozier described a French cigarettefactory in which the maintenance engineershad a considerable say in the plantwide oper-ation. After some probing he discovered thatthe gro'ip possessed the solution to one ofthe major problems faced by the company,that of troubleshooting the elaborate, expen-sive, and irrascible automated machines that

kept breaking down and dumbfoundingeveryone else. It was the one problem thatthe plant manager could in no way control.

The production workers, whiletroublesome from time to time, created noinsurmountable problems; the manager couldreasonably predict their absenteeism or re-place them when necessary. Productionscheduling was something he could deal withsince, by watching inventories and sales, thedemand for cigarettes was known long in ad-vance. Changes in demand could be accom-modated by slowing down or speeding upthe line. Supplies of tobacco and paper werealso easily dealt with through stockpiles andadvance orders.

The one thing that managementcould neither control nor accommodate to,however, was the seemingly happenstancebreakdowns. And the foremen couldn't in-struct the workers what to do when emer-gencies developed since the maintenance de-partment kept its records of problems andsolutiotis locked up in a cabinet or in itsmembers' heads. The breakdowns were, intruth, a critical source of uncertainty for theorganization, and the maintenance engineerswere the only ones who could cope with theproblem.

The engineers' strategic role in cop-ing with breakdowns afl orded them a con-siderable say on plant decisions. Schedulesand production quotas were set in consul-tation with them. And the plant manager,while formally their boss, accepted theirdecisions about personnel in their operation.His submission was to his credit, for withouttheir cooperation he would have had an evenmore difficult time in running the plant.

Ignoring critical consequences

In this cigarette factory, sharing influencewith the maintenance workers reflected the 5

plant manager's awareness of the criticalcontingencies. However, when organization-al members are not aware of the critical con-tingencies they face, and do not share influ-ence accordingly, the failure to do so cancreate havoc. In one case, an insurance com-pany's regional office was having problemswith the performance of one of its depart-ments, the coding department. From the out-side, the department looked like a disasterarea. The clerks who worked in it were some-what dissatisfied; their supervisor paid littleattention to them, and they resented the hardwork. Several other departments were criti-cal of this manager, claiming that she was in-consistent in meeting deadlines. The personmost critical was the claims manager. He re-sented having to wait for work that was han-dled by her department, claiming that it heldup his claims adjusters. Having heard the ru-mors about dissatisfaction among her subor-dinates, he attributed the situation to poorsupervision. He was second in command inthe office and therefore took up the issuewith her immediate boss, the head of admin-istrative services. They consulted with thepersonnel manager and the three of themconcluded that the manager needed leader-ship training to improve her relations withher subordinates. The coding manager ob-jected, saying it was a waste of time, butagreed to go along with the training and alsoagreed to give more priority to the claimsdepartment's work. Within a week after thetraining, the results showed that her workerswere happier but that the performance ofher department had decreased, save for thepeople serving the claims department.

About this time, we began, quiteindependently, a study of influence in thisorganization. We asked the administrativeservices director to draw up flow charts of

6 how the work of one department moved on-

to the next department. In the course of theinterview, we noticed that the coding de-partment began or interceded in the workflow of most of the other departments andcasually mentioned to him, "The codingmanager must be very influential." He said"No, not really. Why would you think so?"Before we could reply he recounted thestory of her leadership training and the factthat things were worse. We then told himthat it seemed obvious that the coding de-partment would be influential from the factthat all the other departments depended onit. It was also clear why productivity hadfallen. The coding manager took the trainingseriously and began spending more time rais-ing her workers' spirits than she did worry-ing about the problems of all the depart-ments that depended on her. Giving priorityto the claims area only exaggerated the prob-lem, for their work was getting done at theexpense of the work of the other depart-ments. Eventually the company hired a fewmore clerks to relieve the pressure in the cod-ing department and performance returned toa more satisfactory level.

Originally we got involved with thisinsurance company to examine how the in-fluence of each manager evolved from his orher department's handling of critical organi-zational contingencies. We reasoned that oneof the most important contingencies faced byall profit-making organizations was that ofgenerating income. Thus we expected man-agers would be influential to the extent towhich they contributed to this function.Such was the case. The underwriting man-agers, who wrote the policies that commit-ted the premiums, were the most influential;the claims managers, who kept a lid on thefunds flowing out, were a close second. Leastinfluential were the managers of functionsunrelated to revenue, such as mailroom and

payroll managers. And contrary to what theadmitiistrative services manager believed,the third most powerful department head(out of 21) was the woman in charge of thecoding function, which consisted of rating,recording, and keeping track of the codes ofall policy applications and contracts. Herpeers attributed more influence to her thancould have been inferred from her place onthe organization chart. And it was not sur-prising, since they all depended on her de-partment. The coding department's records,their accuracy and the speed with whichthey could be retrieved, affected virtuallyevery other operating department in the in-surance office. The underwriters dependedon them in getting the contracts straight; thetyping department depended on them inpreparing the formal contract document;the claims department depended on themin adjusting claims; and accounting depend-ed on them for billing. Unfortunately, the"bosses" were not aware of these depen-dences, for unlike the cigarette factory,there were no massive breakdowns that madethem obvious, while the coding manager,who was a hard-working but quiet person,did little to announce her importance.

The cases of this plant and office il-lustrate nicely a basic point about the sourceof power in orgatiizations. The basis forpower in an organization derives from theability of a person or subunit to take or nottake actions that are desired by others. Thecoding manager was seen as influential bythose who depended on her department, butnot by the people at the top. The engineerswere infiuential because of their role in keep-ing the plant operating. The two cases differin these respects: The coding supervisor'ssource of power was not as widely recog-nized as that of the maintenance engineers,and she did not use her source of power to

influence decisions; the maintenance engi-neers did. Whether power is used to in-fluence anything is a separate issue. Weshould not confuse this issue with the factthat power derives from a social situation inwhich one person has a capacity to do some-thing and another person does not, but wantsit done.

POWER SHARING IN ORGANIZATIONS

Power is shared in organizations; and it isshared out of necessity more than out of con-cern for principles of organizational devel-opment or participatory democracy. Poweris shared because no one person controls allthe desired activities in the organization.While the factory owner may hire people tooperate his noisy machines, once hired theyhave some control over the use of the ma-chinery. And thus they have power overhim in the same way he has power overthem. Who has more power over whom is amooter point than that of recognizing theinherent nature of organizing as a sharing ofpower.

Let's expand on the concept thatpower derives from the activities desired inan organization. A major way of managinginfluence in organizatiotis is through the des-ignation of activities. In a bank we studied,we saw this principle in action. This bankwas planning to install a computer systemfor routine credit evaluation. The bank,rather progressive-minded, was concernedthat the change would have adverse effectson employees and therefore surveyed theirattitudes.

The principal opposition to the newsystem came, interestingly, not from the em-ployees who performed the routine credit 7

checks, some of whom would be relocatedbecause of the change, but from the man-ager of the credit department. His reasonwas quite simple. The manager's primaryfunction was to give official approval to theapplications, catch any employee mistakesbefore giving approval, and arbitrate anydifficulties the clerks had in deciding whatto do. As a consequence of his role, othersin the orgatiization, including his superiors,subordinates, and colleagues, attributed con-siderable imponance to him. He, in turn, forexample, could point to the low proportionof credit approvals, compared with other fi-nancial institutions, that resulted in baddebts. Now, to his mind, a wretched ma-chine threatened to transfer his role to acomputer programmer, a man who knewnothing of finance and who, in addition, hadten years less seniority. The credit managereventually quit for a position at a smaUerfirm with lower pay, but one in which hewould have more influence than his rede-fined job would have left him with.

Because power derives from activi-ties rather than individuals, an individual's orsubgroup's power is never absolute and de-rives ultimately from the context of the sit-uation. The amount of power an individualhas at any one time depends, not only on theactivities he or she controls, but also cn theexistence of other persons or means bywhich the activities can be achieved and onthose who determine what ends are desiredand, hence, on what activities are desired andcritical for the organization. One's ownpower always depends on other people forthese two reasons. Other people, or groupsor orgatiizations, can determine the defini-tion of what is a critical contingency for theorganization and can also undercut theuniqueness of the individual's personal con-tribution to the critical contingencies of theorganization.

Gerald R. Salancik began his studies injournalism, communications, and advertisingat Northwestern University. He laterturned to experimental social psychology,in which he took a doctorate at YaleUniversity in 1910. He has spenta year at the Institute for theFutttre^ where he first studied purposefuladjustment of organizations to environmentalchange. This led naturally to a concern forpower, both within and betweenorganizations. He currently pursues thisinterest as associate professor of organizationbehavior with the Department of BusinessAdministration at the University ofIllinois in Urbana/Champaig7i. He haswritten over 30 papers in the areasof organizational power, coTtrmitment,attitude change, and technologicalforecasting, ayid co-authored The Interview(University of Texas Press, 1966), a7ld tivobooks in pr£jj—New Directions forOrganizational Behavior, to be publishedby St. Clair Press and The External Controlof Organizations, to be published byHarper and Row. He consults on strategicplanning problems for organizations andon issues of control and change.

Perhaps one can best appreciatehow situationally dependent power is by ex-amining how it is distributed. In most so-cieties, power organizes around scarce andcritical resources. Rarely does power or-ganize around abundant resources. In theUnited States, a person doesn't become pow-erful because he or she can drive a car.There are simply too many others who can

drive with equal facility. In certain villagesin Mexico, on the other hand, a person witha car is accredited with enormous socialstatus and plays a key role in the communi-ty. In addition to scarcity, power is alsolimited by the need for one's capacities ina social system. While a racer's ability todrive a car around a 90° tum at 80 mph maybe sparsely distributed in a society, it is notlikely to lend the driver much power in thesociety. The ability simply does not play acentral role in the activities of the society.

The fact that power revolvesaround scarce and critical activities, ofcourse, makes the control and organizationof those activities a major battleground instruggles for power. Even relatively abun-dant or trivial resources can become thebases for power if one can organize and con-trol their allocation and the definition ofwhat is critical. Many occupational and pro-fessional groups attempt to do just this inmodem economies. Lawyers organize them-selves into associations, regulate the entrancerequirements for novitiates, and then getlaws passed specifying situations that requirethe services of an attorney. Workers hadlittle power in the conduct of industrial af-fairs until they organized themselves intoclosed and controlled systems. In recentyears, woijien and blacks have tried to de-fine themselves as important and critical tothe social system, using law to reify theirstatus.

In organizations there are obvious-ly opportunities for defining certain activi-ties as more critical than others. Indeed, thegrowth of managerial thinking to include de-fining organizational objectives and goals hasdone much to foster these opportunities. Onesure way to liquidate the power of groups inthe organization is to define the need for theirservices out of existence. David Halberstampresents a description of how just such a

Jeffrey Pfeffer received his ?hD. from theGraduate School of Business at StanfordUniversity and obtained BS. and MS.degrees from the Graduate School ofIndustrial Adrninistration at Carnegie-MellonUniversity. He taught at the University ofIllinois at Urbana-Champaign beforecoming to the University of California atBerkeley, ivhere he is currently an associateprofessor in the School of BusinessAdministration and is also affiliated withthe institute of Industrial Relations. Hisresearch has dealt with the topics oforganizational structure and design, powerand decision making in organizations, andthe strategic actions taken by organizationsto manage their resource dependence onthe social environment. Most recently,he has been working on a project touse power-dependence concepts to explainsupervisory behavior and effectivenessand to examine the process by whichconsequences are attributed to manageri^dactions.

Professor Pfeffer has consulted in theareas of decision making and organizationaland management development, and haspublished extenswely m the areas of power inorganizations and environmental constraintsand organizational responses. Along withGerald Salancik, he has written The ExternalControl of Organizations: A ResourceDependence Perspective, to be published byHarper and Row.

thing happened to the group of correspon-dents that evolved around Edward R. Mur-row, the brilliant journalist, interviewer, andwar correspondent of CBS News. A close 9

friend of CBS chairman and controllingstockholder William S. Paley, Murrow, andthe news department he directed, were en-dowed with freedom to do what they feltwas right. He used it to create some of thebest documentaries and commentaries everseen on television. Unfortunately, televisionbecame too large, too powerful, and too sus-pect in the eyes of the federal governmentthat licensed it. It thus became, or at least thetop executives believed it had become, toodangerous to have in-depth, probing com-mentary on the news. Crisp, dry, uneditorial-izing headliners were considered safer. Mur-row was out and Walter Cronkite was in.

The power to define what is criticalin an organization is no small power. More-over, it is the key to understanding why or-ganizations are either aligned with their en-vironments or misaligned. If an organizationdefines certain activities as critical when infact they are not critical, given the flow ofresources coming into the organization, it isnot likely to survive, at least in its presentform.

Most organizations manage toevolve a distribution of power and influencethat is aligned with the critical realities theyface in the environment. The environment,in turn, includes both the internal environ-ment, the shifting situational contexts inwhich particular decisions get made, andthe external environment that it can hopeto influence but is unlikely to control.

THE CRITICAL CONTINGENCIES

The critical contingencies facing most or-ganizations derive from the environmentalcontext within which they operate. This de-termines the available needed resources andthus determines the problems to be dealtwith. That power organizes around han-

10 dling these problems suggests an important

mechanism by which organizations keep intune with their external environments. Thestrategic-contingencies model implies thatsubunits that contribute to the critical re-sources of the organization will gain influ-ence in the organization. Their influencepresumably is then used to bend the organi-zation's activities to the contingencies thatdetermine its resources. This idea maystrike one as obvious. But its obviousness inno way diminishes its importance. Indeed,despite its obviousness, it escapes the noticeof many organizational analysts and manag-ers, who all too frequently think of the or-ganization in terms of a descending pyramid,in which all the departments in one tier holdequal power and status. This presumptiondenies the reality that departments diflFerin the contributions they are believed tomake to the overall organization's resources,as well as to the fact that some are moreequal than others.

Because of the importance of thisidea to organizational effectiveness, we de-cided to examine it carefully in a large mid-western university. A university offers anexcellent site for studying power. It is com-posed of departments with nominally equalpower and is administered by a central ex-ecutive structure much like other bureaucra-cies. However, at the same time it is a situ-ation in which the departments have clearlydefined identities and face diverse externalenvironments. Each department has its ownbodies of knowledge, its own institutions,its own sources of prestige and resources.Because the departments operate in differ-ent external environments, they are likely tocontribute differentially to the resources ofthe overall organization. Thus a physics de-partment with close ties to NASA may con-tribute substantially to the funds of the uni-versity; and a history depanment with arenowned historian in residence may con-

tribute to the intellectual credibility or pres-tige of the whole university. Such variationspermit one to examine how these variouscontributions lead to obtaining power with-in the university.

We analyzed the influence of 29university departments throughout an 18-month period in their history. Our chief in-terest was to determine whether departmentsthat brought more critical resources to theuniversity would be more powerful than de-partments that contributed fewer or less crit-ical resources.

To identify the critical resourceseach department contributed, the heads ofall departments were interviewed about theimponance of seven different resourcesto the university's success. The seven in-cluded undergraduate students (the factordetermining size of the state allocations bythe university), national prestige, administra-tive expertise, and so on. The most criticalresource was found to be contract and grantmonies received by a department's facultyfor research or consulting services. At thisuniversity, contract and grants contributedsomewhat less than 50 percent of the over-all budget, with the remainder primarilycoming from state appropriations. The im-portance attributed to contract and grantmonies, and the rather minor importance ofundergraduate students, was not surprisingfor this particular university. The universitywas a major center for graduate education;many of its departments ranked in the topten of their respective fields. Grant and con-tract monies were the primary source ofdiscretionary funding available for main-taining these programs of graduate educa-tion, and hence for maintaining the univer-sity's prestige. The prestige of the universityitself was critical both in recruiting able stu-dents and attracting top-notch faculty.

From university records it was de-

termined what relative contributions each ofthe 29 departments made to the variousneeds of the university (national prestige,outside grants, teaching). Thus, for instance,one department may have contributed tothe university by teaching 7 percent of theinstructional units, bringing in 2 percent ofthe outside contracts and grants, and havinga national ranking of 20. Another depart-ment, on the other hand, may have taughtone percent of the instructional units, con-tributed 12 percent to the grants, and beranked the third best department in its fieldwithin the country.

The question was: Do these differ-ent contributions determine the relativepower of the departments within the uni-versity? Power was measured in severalways; but regardless of how measured, theanswer was "Yes." Those three resourcestogether accounted for about 70 percent ofthe variance in subunit power in the uni-versity.

But the most important predictorof departmental power was the department'scontribution to the contracts and grants ofthe university. Sixty percent of the variancein power was due to this one factor, suggest-ing that the power of departments derivedprimarily from the dollars they provided forgraduate education, the activity believed tobe the most important for the organization.

THE IMPACT OF ORGANIZATIONAL POWER

ON DECISION MAKING

The measure of power we used in studyingthis university was an analysis of the re-sponses of the department heads we inter-viewed. While such perceptions of powermight be of interest in their own right, theycontribute little to our understanding ofhow the distribution of power might serveto align an organization with its critical re- 11

alities. For this we must look to how poweractually influences the decisions and policiesof organizations.

While it is perhaps not absolutelyvalid, we can generally gauge the relativeimportance of a department of an organiza-tion by the size of the budget allocated toit relative to other departments. Clearly it isof importance to the administrators of thosedepartments whether they get squeezed ina budget crunch or are given more funds tostrike out after new opportunities. And itshould also be clear that when those deci-sions are made and one department can goahead and try new approaches while anothermust cut back on the old, then the deploy-ment of the resources of the organization inmeeting its problems is most directly af-fected.

Thus our study of the universityled us to ask the following question: Doespower lead to influence in the organization?To answer this question, we found it usefulfirst to ask another one, namely: Whyshould department heads try to influenceorganizational decisions to favor their owndepartments to the exclusion of other de-partments? While this second question mayseem a bit naive to anyone who has wit-nessed the political realities of organizations,we posed it in a context of research on orga-nizations that sees power as an illegitimatethreat to the neater rational authority ofmodem bureaucracies. In this context, de-cisions are not believed to be made becauseof the dirty business of politics but becauseof the overall goals and purposes of the or-ganization. In a university, one reasonablebasis for decision making is the teachingworkload of departments and the demandsthat follow from that workload. We wouldexpect, therefore, that departments withheavy student demands for courses would

12 be able to obtain funds for teaching. An-

other reasonable basis for decision making isquality. We would expect, for that reason,that departments with esteemed reputationswould be able to obtain funds both becausetheir quality suggests they might use suchfunds effectively and because such fundswould allow them to maintain their quality.A rational model of bureaucracy intimates,then, that the organizational decisions takenwould favor those who perform the statedpurposes of the organization—teaching un-dergraduates and training professional andscientific talent—well.

The problem with rational modelsof decision making, however, is that what isrational to one person may strike another asirrational. For most departments, resourcesare a question of survival. While teachingundergraduates may seem to be a major goalfor some members of the university, devel-oping knowledge may seem so to others; andto still others, advising governments andother institutions about policies may seem tobe the crucial business. Everyone has his ownidea of the proper priorities in a just world.Thus goals rather than being clearly definedand universally agreed upon are blurred andcontested throughout the organization. Ifsuch is the case, then the decisions taken onbehalf of the organization as a whole arelikely to reflect the goals of those who pre-vail in political contests, namely, those withpower in the organization.

Will organizational decisions al-ways reflect the distribution of power in theorganization? Probably not. Using powerfor influence requires a certain expenditureof effort, time, and resources. Prudent andjudicious persons are not likely to use theirpower needlessly or wastefuUy. And it islikely that power will be used to influenceorganizational decisions primarily under cir-cumstances that both require and favor itsuse. We have examined three conditions that

are likely to affect the use of power in or-ganizations: scarcity, cridcality, and uncer-tainty. The first suggests that subunits willtry to exert infiuence when the resources ofthe organization are scarce. If there is anabundance of resources, then a particular de-partment or a particular individual has littleneed to attempt infiuence. With little efifort,he can get all he wants anyway.

The second condition, criticality,suggests that a subunit will attempt to in-fluence decisions to obtain resources that arecritical to its own survival and activities.Criticality impUes that one would not wasteeffort, or risk being labeled obstinate, byfighting over trivial decisions affecting one'soperations.

An ofiSce manager would probablybalk less about a threatened cutback in copy-ing machine usage than about a reduction intyping staff. An advertising department headwould probably worry less about losing hislettering anist than his illustrator. Criticalityis difficult to define because what is criticaldepends on people's beliefs about what iscritical. Such beliefs may or may not be basedon experience and knowledge and may ormay not be agreed upon by all. Scarcity,for instance, may itself affect conceptionsof criticality. When slack resources dropoff, cutbacks have to be made—those "harddecisions," as congressmen and resplendentadministrators like to call them. Managersthen find themselves scrapping projects theyonce held dear.

The third condition that we believeaffects the use of power is uncertainty:When individuals do not agree about whatthe organization should do or how to do it»power and other social processes will affectdecisions. The reason for this is simply that,if there are no clear-cut criteria available forresolving conflicts of interest, then the onlymeans for resolution is some form of social

process, including power, status, social des,or some arbitrary process like flipping a coinor drawing suaws. Under condidons of un-certainty, the powerful manager can arguehis case on any grounds and usually win it.Since there is no real consensus, other con-testants are not likely to develop counterarguments or amass sufiicient opposition.Moreover, because of his power and theirneed for access to the resources he controls,they are more likely to defer to his argu-ments.

Although the evidence is slight, wehave found that power will influence the al-locations of scarce and critical resources. Inthe analysis of power in the university, forinstance, one of the most cridcal resourcesneeded by departments is the general budget.First granted by the state legislature, the gen-eral budget is later allocated to individual de-partments by the university administradonin response to requests from the departmentheads. Our analysis of the factors that con-tribute to a department getdng more or lessof this budget indicated that subunit powerwas the major predictor, overriding suchfactors as student demand for courses, na-tional reputadons of departments, or eventhe size of a department's faculty. Moreover,other research has shown that when the gen-eral budget has been cut back or held belowprevious uninflated levels, leading to moniesbecoming more scarce, budget allocadonsmirror departmental powers even moreclosely.

Student enrollment and faculty size,of course, do themselves relate to budgetallocadons, as we would expect since theydetermine a department's need for resources,or at least offer visible tesdmony of needs.But departments are not always able toget what they need by the mere fact ofneeding them. In one analysis it was foundthat high-power departments were able to 13

obtain budget without regard to their teach-ing loads and, in some cases, actually in in-verse relation to their teaching loads. In con-trast, low-power departments could get in-creases in budget only when they couldjustify the increases by a recent growth inteaching load, and then only when it wasfar in excess of norms for other departments.

General budget is only one form ofresource that is allocated to departments.There are others such as special grants forstudent fellowships or faculty research.These are critical to departments becausethey affect the ability to attract other re-sources, such as outstanding faculty or stu-dents. We examined how power influencedthe allocations of four resources departmentbeads had described as critical and scarce.

When the four resources were ar-rayed from the most to the least critical andscarce, we found that departmental powerbest predicted the allocations of the mostcritical and scarce resources. In other words,the analysis of how power influences organi-zational allocations leads to this conclusion:Those subunits most likely to survive intimes of strife are those that are more criticalto the organization. Their importance to theorganization gives them power to influenceresource allocations that enhance their ownsurvival.

How EXTERNAL ENVIRONMENT

IMPACTS EXECUTIVE SELECTION

Power not only influences the survival ofkey groups in an organization, it also in-fluences the selection of individuals to keyleadership positions, and by such a processfurther aligns the organization with its en-vironmental context.

14 We can illustrate this with a recent

study of the selection and tenure of chiefadministrators in 57 hospitals in Illinois. Weassumed that since the critical problems fac-ing the organization would enhance thepower of certain groups at the expense ofothers, then the leaders to emerge should bethose most relevant to the context of thehospitals. To assess this we asked each chiefadministrator about his professional back-ground and how long he had been in office.The replies were then related to the hos-pitals' funding, ownership, and competitiveconditions for patients and staff.

One aspect of a hospital's context isthe source of its budget. Some hospitals, forinstance, are run much like other businesses.They sell bed space, patient care, and treat-ment services. They charge fees sufficientboth to cover their costs and to provide cap-ital for expansion. The main source of boththeir operating and capital funds is patientbillings. Increasingly, patient billings are paidfor, not by patients, but by private insurancecompanies. Insurers like Blue Cross domi-nate and represent a potent interest groupoutside a hospital's control but critical to itsincome. The insurance companies, in orderto limit their own costs, attempt to holddown the fees allowable to hospitals, whichthey do effectively from their positions onstate rate boards. The squeeze on hospitalsthat results from fees increasing slowly whilecosts climb rapidly more and more demandsthe talents of cost accountants or peopletrained in the technical expertise of hospitaladministration.

By contrast, other hospitals operatemore like social service institutions, eitheras government healthcare units (BellevueHospital in New York City and Cook Coun-ty Hospital in Chicago, for example) or ascharitable institutions. These hospitals ob-tain a large proportion of their operating

and capital funds, not from privately in-sured patients, but from government sub-sidies or private donations. Such institutionsrather than requiring the talents of a techni-cally efficient administrator are likely to re-quire the savvy of someone who is well inte-grated into the social and political powerstructure of the community.

Not surprisingly, the characteristicsof administrators predictably reflect thefunding context of the hospitals with whichthey arc associated. Those hospitals withlarger proportions of their budget obtainedfrom private insurance companies were mostlikely to have administrators with back-grounds in accounting and least likely tohave administrators whose professions werebusiness or medicine. In contrast, those hos-pitals with larger proportions of their budgetderived from private donations and localgovernments were most likely to have ad-ministrators with business or professionalbackgrounds and least likely to have accoun-tants. TTie same held for formal training inhospital management. Professional hospitaladministrators could easily be found in hos-pitals drawing their incomes from privateinsurance and rarely in hospitals dependenton donations or legislative appropriations.

As with the selection of administra-tors, the context of organizations has alsobeen found to affect the removal of execu-tives. The environment, as a source of or-ganizational problems, can make it more orless difficult for executives to demonstratetheir value to the organization. In the hospi-tals we studied, long-term administratorscame from hospitals with few problems.They enjoyed amicable and stable relationswith their local business and social commu-nities and suffered little competition forfunding and staff. The small city hospital di-rector who attended civic and Elks meetings

while running the only hospital within a 100-mile radius, for example, had little difficultyholding on to his job. Turnover was highestin hospitals with the most problems, a phen-nomenon similar to that observed in a studyof industrial organizations in which turn-over was highest among executives in indus-tries with competitive environments and un-stable market conditions. The interestingthing is that instability characterized the in-dustries rather than the individual firms inthem. The troublesome conditions in the in-dividual firms were attributed, or rather mis-attributed, to the executives themselves.

It takes more than problems, how-ever, to terminate a manager's leadership.The problems themselves must be relevantand critical. This is clear from the way inwhich an administrator's tenure is affectedby the status of tbe hospital's operating bud-get. Naively we might assume that all ad-ministrators would need to show a surplus.Not necessarly so. Again, we must distin-guish between those hospitals that dependon private donations for funds and those thatdo not. Whether an endowed budget showsa surplus or deficit is less important than thehospital's relations with benefactors. On theother hand, with a budget dependent onpatient billing, a surplus is almost essential;monies for new equipment or expansionmust be drawn from it, and without themquality care becomes more diflScult and pa-tients scarcer. An administrator's tenure re-flected just these considerations. For thosehospitals dependent upon private donations,the length of an administrator's term de-pended not at all on the status of the oper-ating budget but was fairly predictable fromthe hospital's relations with the business com-munity. On the other hand, in hospitals de-pendent on rhe operating budget for capitalfinancing, the greater the deficit the shorter 15

was the tenure of the hospital's principal ad-ministrators.

CHANGING CONTINGENCIES AND

ERODING POWER BASES

The critical contingencies facing the orga-nization may change. When they do, it isreasonable to expect that the power of in-dividuals and subgroups will change in tum.At times the shift can be swift and shatter-ing, as it was recently for powerholders inNew York Gty. A few years ago it was be-lieved that David Rockefeller was one of theten most powerful people in the city, astallied by New York magazine, which an-nually sniffs out power for the delectationof its readers. But that was before it was re-vealed that the city was in financial trouble,before Rockefeller's Chase Manhattan Banklost some of its own financial luster, and be-fore brother Nelson lost some of his politicalinfluence in Washington. Obviously DavidRockefeller was no longer as well positionedto help bail the city out. Another loser wasan attorney with considerable personal con-nections to the political and religious leadersof the city. His talents were no longer inmuch demand. The persons with more in-fluence were the bankers and union pensionfund executors who fed money to the city;community leaders who represent blacks andSpanish-Americans, in contrast, witnessedthe erosion of their power bases.

One implication of the idea thatpower shifts with changes in organizationalenvironments is that the dominant coalitionwill tend to be that group that is most ap-propriate for the organizatioa's environ-ment, as also will the leaders of an organi-zation. One can observe this historically inthe top executives of industrial firms in

16 the United States. Up until the early 1950s,

many top corporations were headed byformer production line managers or engi-neers who gained prominence because oftheir abilities to cope with the problems ofproduction. Their success, however, onlyspelled their demise. As production becameroutinized and mechanized, the problem ofmost firms became one of selling all thosegoods they so efficiently produced. Market-ing executives were more frequently foundin corporate boardrooms. Success outdid it-self again, for keeping markets and produc-tion steady and stable requires the kind ofcontrol that can only come from acquiringcompetitors and suppliers or the inventionof more and more appealing products—ventures that typically require enormousamounts of capital. During the 1960s, finan-cial executives assumed the seats of power.And they, too, will give way to others. Edg-ing over the horizon are legal experts, asregulation and antitrust suits are becomingmore and more frequent in the 1970s, suitsthat had their beginnings in the success ofthe expansion generated by prior executives.The more distant future, which is likely tobe dominated by multinational corporations,may see former secretaries of state and theirminions increasingly serving as corporatefigureheads.

THE NONADAPTIVE CONSEQUENCES

OF ADAPTATION

From what we have said thus far aboutpower aligning the organization with its ownrealities, an intelligent person might reactwith a resounding ho-hum, for it all seemstoo obvious: Those with the ability to get thejob done are given the job to do.

However, there are two aspects ofpower that make it more useful for under-standing organizations and their effective-ness. First, the "job" to be done has a wayof expanding itself until it becomes less andless clear what the job is. Napoleon beganby doing a job for France in the war withAustria and ended up Emperor, convincingmany that only he could keep the peace.Hitler began by promising an end to Ger-many's troubling postwar depression andended up convincing more people than iscomfortable to remember thar he was des-tined to be the savior of the world. In short,power is a capacity for influence that ex-tends far beyond the original bases thatcreated it. Second, power tends to take oninstitutionalized forms that enable it to en-dure weU beyond its usefulness to an organi-zation.

There is an important contradic-tion in what we have observed about orga-nizational power. On the one hand we havesaid that power derives from the contingen-cies facing an organization and that whenthose contingencies change so do the basesfor power. On the other hand we have as-serted that subunits will tend to use theirpower to influence organizational decisionsin their own favor, particularly when theirown survival is threatened by the scarcity ofcritical resources. The first statement im-plies that an organization will tend to bealigned with its environment since powerwill tend to bring to key positions thosewith capabilities relevant to the context. Thesecond implies that those in power will notgive up their positions so easily; they willpursue policies that guarantee their contin-ued domination. In short, change and stabil-ity operate through the same mechanism,

and, as a result, the organization will neverbe completely in phase with its environmentor its needs.

The study of hospital administra-tors illustrates how leadership can be out ofphase with reality. We argued that privatelyfunded hospitals needed trained technical ad-ministrators more so than did hospitals fund-ed by donations. The need as we perceivedit was matched in most hospitals, but byno means in all. Some organizations did notconform with our predictions. These devia-tions imply that some administrators wereable to maintain their positions independentof their suitability for those positions. Bydividing administrators into those with longand shon terms of office, one finds that thecharacteristics of longer-termed administra-tors were virtually unrelated to the hospi-tal's context. The shorter-termed chiefs onthe other hand had characteristics more ap-propriate for the hospital's problems. For ahospital to have a recently appointed headimplies that the previous administrator hadbeen unable to endure by institutionalizinghimself.

One obvious feature of hospitalsthat allowed some administrators to enjoy along tenure was a hospital's ownership. Ad-ministrators were less entrenched when theirhospitals were affiliated with and dependentupon larger organizations, such as govern-ments or churches. Private hospitals offeredmore secure positions for administrators.Like private corporations, they tend to havemore diffused ownership, leaving the admin-istrator unopposed as he institutionalizes hisreign. Thus he endures, sometimes at the ex-pense of the performance of the organization.Other research has demonstrated that cor-porations with diffuse ownership have poor-er earnings than those in which the controlof the manager is checked by a dominantshareholder. Firms that overload their board- 17

rooms with more insiders than are appro-priate for their context have also been foundto be less profitable.

A word of caution is requiredabout our judgment of "appropriateness."When we argue some capabilities are moreappropriate for one context than another, wedo so from the perspective of an outsiderand on the basis of reasonable assumptions asto the problems the organization will faceand the capabilities they will need. The factthat we have been able to predict the distri-bution of influence and the characteristics ofleaders suggests that our reasoning is not in-correct. However, we do not think that allorganizations follow the same pattern. Thefact that we have not been able to predictoutcomes with 100 percent accuracy indi-cates they do not.

MISTAKING CRITICAL CONTINGENCIES

One thing that allows subunits to retain theirpower is their ability to name their functionsas critical to the organization when they maynot be. Consider again our discussion ofpower in the university. One might wonderwhy the most critical tasks were de-fined as graduate education and scholarlyresearch, the effect of which was to lendpower to those who brought in grants andcontracts. Why not something else? Thereason is that the more powerful depart-ments argued for those criteria and wontheir case, partly because they were morepowerful.

In another analysis of this universi-ty, we found that all departments advocateself-serving criteria for budget allocation.Thus a department with large undergrad-uate enrollments argued that enrollmentsshould determine budget allocations, a de-

18 partment with a strong national reputation

saw prestige as the most reasonable basis fordistributing funds, and so on. We furtherfound that advocating such self-serving cri-teria actually benefited a department's bud-get allotments but, also, it paid off more fordepartments that were already powerful.

Organizational needs are consistentwith a current distribution of power alsobecause of a human tendency to categorizeproblems in familiar ways. An accountantsees problems with organizational perfor-mance as cost accountancy problems or in-ventory flow problems. A sales manager seesthem as problems with markets, promotionalstrategies, or just unaggressive salespeople.But what is the truth? Since it does not auto-matically announce itself, it is likely thatthose with prior credibility, or those withpower, will be favored as the enlightened.This bias, while not intentionally self-serv-ing, further concentrates power among thosewho already possess it, independent ofchanges in the organization's context.

INSTITUTIONALIZING POWER

A third reason for expecting organizadonalcontingencies to be defined in familiar waysis that the current holders of power canstructure the organization in ways that in-stitutionalize themselves. By institutionaliza-tion we mean the establishment of relativelypermanent structures and policies that favorthe influence of a particular subunit. Whilein power, a dominant coalition has the abil-ity to institute constitutions, rules, proce-dures, and information systems that limit thepotential power of others while continuingtheir own.

The key to institutionalizing poweralways is to create a device that legitimatesone's own authority and diminishes the le-gitimacy of others. When the "Divine Right

of Kings" was envisioned centuries ago itwas to provide an unquestionable foutidationfor the supremacy of royal authority. Thereis generally a need to root the exercise ofauthority in some higher power. Modemleaders are no less affected by this need.Richard Nixon, with the aid of John Dean,reified the concept of executive privilege,which meant in effect that what the Presi-dent wished not to be discussed need notbe discussed.

In its simpler form, institutionaliza-don is achieved by designating positions orroles for organizational activities. The cre-ation of a new post legitimizes a functionand forces organization members to orientto it. By designating how this new post re-ktes to older, more established posts, more-over, one can structure an organization toenhance the importance of the function inthe organization. Equally, one can diminishthe importance of traditional functions. Thisis what happened in the end with the in-surance company we mentioned that washaving trouble with its coding department.As the situation unfolded, the claims direc-tor continued to feel dissatisfied about thedependency of his functions on the codingmanager. Thus he instituted a reorganizationthat resulted in two coding departments. Inso doing, of course, he placed activities thataffected his department under his direct con-trol, presumably to make the operation moreeffective. Similarly, consumer-product firmsenhance the power of marketing by settingup a coordinating role to interface produc-tion and marketing functions and then ap-point a marketing manager to fill the role.

The structures created by dominantpowers sooner or later become fixed and un-questioned features of the organization.Eventually, this can be devastating. It is saidthat the battle of Jena in 1806 was lost byFrederick the Great, who died in 1786.

Though the great Prussian leader had nodirect hand in the disaster, his imprint on thearmy was so thorough, so embedded in itsskeletal underpinnings, that the organizationwas inappropriate for others to lead in dif-ferent times.

Another important source of insti-tutionalized pow er lies in the ability tostructure information systems. Setting upcommittees to investigate particular organi-zational issues and having them repon onlyto particular individuals or groups, facilitatestheir awareness of problems by members ofthose groups while limiting the awareness ofproblems by the members of other groups.Obviously, those who have information arein a better position to interpret the problemsof an organization, regardless of how realis-tically they may, in fact, do so.

Still another way to institutionalizepower is to distribute rewards and resources.The dominant group may quiet competinginterest groups with small favors and re-wards. The credit for this artful form of co-optation belongs to Louis XIV. To avoidusurpation of his power by the nobles ofFrance and the Fronde that had so troubledhis father's reign, he built the palace at Ver-sailles to occupy them with hunting and gos-sip. Awed, the courtiers basked in the reflect-ed glories of the "Sun King" and the over-whelming setting he had created for hiscourt.

At this point, we have not system-atically studied the institutionalization ofpower. But we suspect it is an importantcondition that mediates between the envi-ronment of the organization and the capabil-ities of the organization for dealing withthat environment. The more institutional-ized power is within an organization, themore likely an organization will be out ofphase with the realities it faces. PresidentRichard Nixon's structuring of his White 19

House is one of the better documented il-lustrations. If we go back to tiewspaper andmagazine descriptions of how he organizedhis office from the beginning in 1968, mostof what occurred subsequently follows al-most as an afterthought. Decisiotis flowedthrough virtually only the smali WhiteHouse staff; rewards, small presidential fav-ors of recogtiidon, and perquisites were dis-tributed by this staff to the loyal; atid itifor-mation from the outside world—the press,Congress, the people on the streets—wasfiltered by the staff and passed along only ifinitialed "bh." Thus it was not surprising thatwhen Nixon met war protestors in the earlydawn, the otily thing he could think to talkabout was the latest football game, so itisu-lated had he become from their grief andanger.

One of the more interesting impli-cations of institutionalized power is that ex-ecutive turnover among the executives whohave structured the organization is likelyto be a rare event that occurs only under themost pressing crisis. If a dominant coalitionis able to structure the organization and in-terpret the meaning of ambiguous eventslike declining sales and profits or lawsuits,then the "real" problems to emerge will eas-ily be incorporated into traditional molds ofthinking and acting. If opposition is designedout of the orgatiization, the interpretationswill go unquestioned. Conditions will re-main stable until a crisis develops, so over-whelming and visible that even the mostadroit rhetorician would be silenced.

IMPLICATIONS FOR THE MANAGEMENT

OF POWER IN ORGANIZATIONS

While we could derive numerous implica-tions from this discussion of power, our se-

20 lection would have to depend largely on

whether one wanted to increase one's power,decrease the power of others, or merelymaintain one's position. More important, thereal implications depend on the particularsof an organizational situation. To under-stand power in an organization one must be-gin by looking outside it—into the environ-ment—for those groups that mediate the or-ganization's outcomes but are not themselveswithin its control.

Instead of ending with homilies, wewill end with a reversal of where we began.Power, rather than being the dirty businessit is often made out to be, is probably one ofthe few mechanisms for reality testing in or-ganizations. And the cleaner forms of power,the institutional forms, rather than havingthe virtues they are often credited with, canlead the organization to become out oftouch. The real trick to managing power inorganizations is to ensure somehow that lead-ers cannot be unaware of the realities of theirenvironments and cannot avoid changing todeal with those realities. That, however,would be like designing the "self-liquidatingorganization," an unlikely event since any-one capable of designing such an instrumentwould be obviously in control of the liqui-dations.

Management would do well to de-vote more attention to determitiing the crit-ical contingencies of their environments.For if you conclude, as we do, that the en-vironment sets most of the structure influ-encing organizational outcomes and prob-lems, and that power derives from the orga-nization's activities that deal with those con-tingencies, then it is the environment thatneeds managing, not power. The first stepis to construct an accurate model of the en-vironment, a process that is quite difficultfor most organizations. We have recentlystarted a project to aid adtninistrators insystematically understanding their environ-

ments. From this experience, we havelearned that the most critical blockage to per-ceiving an organization's reality accurately isa failure to incorporate those with the rele-vant expertise into the process. Most organi-zations have the requisite experts on hatid butthey arc positioned so that they can be com-fortably ignored.

One conclusion you can, and prob-

ably should, derive from our discussion isthat power—because of the way it developsand the way it is used—will always result inthe organization suboptimizing its perfor-mance. However, to this grim absolute, weadd a comforting caveat: If any criteriaother than power were the basis for deter-mining an organization's decisions, the re-sults would be even worse.

SEI£CTED BIBLIOGRAPHY

The literature on power is at once both volumi-nous and frequently empty of content. Some isphilosophical musing about the concept ofpower, while other writing contains popular-ized palliatives for acquiring and exercising in-fiuence. Machiavelli's The Prince, if read care-fully, remains the single best prescriptive treat-ment of power and its use. Most social scientistshave approached power descriptively, attempt-ing to understand how it is acquired, how it isused, and what its effects are. Mayer Zald's edi-ted collection Poauer in Organizations (Vander-bilt University Press, 1970) is one of the moreuseful sets of thoughts about power from a so-ciological perspective, while James Tedeschi'sedited book. The Social Influence Processes(Aldine-Atherton, 1972) represents the socialpsychological approach to understanding powerand influence. The strategic contingencies's ap-proach, with its emphasis on the importance ofuncertainty for understanding power in organi-zations, is described by David Hickson and hiscolleagues in "A Strategic Contingencies The-ory of Intraorganizational Power" {Administra-tive Science Quarterly^ December 1971, pp.216-229).

Unfortunately, while many have writ-

ten about power theoretically, there have beenfew empirical examinations of power and its use.Most of the work has taken the form of casestudies. Michel Crozier's The Bureaucratic Phe-nomenon (University of Chicago Press, 1964)is important because it describes a group's sourceof power as control over critical activities andillustrates how power is not strictly derivedfrom hierarchical position. J. Victor Baldridge'sPower and Conflict in the University (JohnWiley & Sons, 1971) and Andrew Pettigrew'sstudy of computer purchase decisions in oneEnglish firm (Politics of Organizational Deci-sion-Makings Tavistock, 1973) both present in-sights into the acquisition and use of power inspecific instances. Our work has been more em-pirical and comparative, testing more explicitlythe ideas presented in this article. The study ofuniversity decision making is reported in articlesin the June 1974, pp. 135-151, and December1974, pp. 453-473, issues of the AdministrativeScience Quarterly, the insurance firm study inJ. G. Hunt and L. L. Larson's collection^Leadership Frontiers (Kent State UniversityPress, 1975), and the study of hospital adminis-trator succession will appear in 1977 in theAcademy of Management journal. 21

©1977 Elsevier