7
Choosing Control and Improving Control

Close inspection has revealed that democratic control of bureaucracy is neither analytically simple nor easy to achieve. Yet the goal is critically important for achieving a truly democratic government. In this chapter I focus on three related lessons from this research: the need to think about control in a differentiated fashion, the need to understand the broader context in which control is exercised, and the need to adopt a strategic perspective on the task of controlling bureaucracies.

The Many Faces of Control

Everybody complains about uncontrolled bureaucracies, and many think that they can do something about them. What that something is, however, varies considerably. One of my major tasks in this book has been to illuminate this variation. Democratic control of bureaucracy has many faces. It is a school board instructing the superintendent to emphasize basic skills in the elementary grades; it is a mayor instructing a fire chief to decide which station to close only after holding public hearings; it is a law requiring that housing authorities work with

tenant councils to develop new maintenance policies; it is Congress instructing the secretary of agriculture as to who shall be eligible for food stamps; it is the president requesting the National Endowment for the Arts to avoid concentrating its grants in big cities. There are many different ways to reconcile bureaucracy with democracy. Attention to only one will blind would-be controllers to the complexity of democratic control. Controllers must decide how they will exercise control, what values they will pursue, and what costs they will tolerate. They must also make these choices in the context of particular policy arenas that differ from one another in important ways.

First, controllers must choose among multiple means of constraining bureaucratic behavior. One set of choices involves the question of whether the limits placed on acceptable behavior involve the procedures bureaucrats use to make decisions or the actual substance of those decisions. In the first case control is exercised over the way bureaucrats decide, as the mayor does by requiring the fire chief to hold hearings. In the second case, as when the school board directs the superintendent to emphasize basic skills, control is exercised over what those decisions are.

A second choice involves the extent to which bureaucrats are constrained. Controls run the gamut from loose, ruling out a very limited set of options, to tight, ruling in an equally limited set. The president is exerting loose control over the National Endowment for the Arts when he tells it to avoid concentrating grants. Congress tightly constrains the Department of Agriculture when it specifies precise eligibility requirements for food stamps.

Controllers must also face a set of choices concerning their own beliefs about democracy. Control over crucial policy-making institutions such as bureaucracies is integral to the proper functioning of a democracy. Since the

content of the idea of a properly functioning democracy is itself widely disputed, however, would-be controllers must choose which democratic beliefs they will predicate control upon. They must come to grips with how capable they think citizens are of governing themselves, and, more specifically, with how capable citizens and their elected officials are compared to bureaucrats. They must also position themselves between, two contrasting reasons for having a democratic government—one primarily liberty-seeking, the other action-seeking. The first emphasizes the protection of the rights of the individual; the second is primarily concerned with the state as a vehicle for achieving the will of the majority of its citizens.

Controlling bureaucracies is rarely a costless enterprise. Controllers must balance the benefits of control against at least three kinds of costs. First, constraining the range of acceptable bureaucratic behavior will at times impair the ability of an agency to carry out its mission. The bureaucracy will fail to attain its original objectives as well as it otherwise might; constraint creates a cost to bureaucratic effectiveness. Second, controllers may find it difficult to determine whether bureaucrats have made a good faith effort to act within the bounds of a constraint. Under these circumstances, they may be reluctant to impose sanctions because they do not want to penalize the bureaucrats for conditions beyond their control. As a result there will be a cost in the controllers' ability to enforce constraint. Controllers who do enforce the constraint may create resentment (or worse) among bureaucrats who think they are being penalized for things beyond their control. Third, controllers frequently encounter resistance from the bureaucrats themselves. When this happens control will be difficult to bring about and may exact a heavy cost to the controller's resources. In the extreme, the controller may abandon the attempt

at constraint in order to husband resources for other uses.

A differentiated understanding of the problem of democratic control is not, however, necessarily an unsystematic one. Although this may seem to be a daunting set of choices, a fundamental logic links different kinds of constraint with different democratic values and different costs.

Democratic beliefs are linked to different kinds of constraint through a coherence of vision. Tight constraint of bureaucratic behavior presumes that citizens or their elected officials are capable of making detailed decisions about the work of a bureaucracy. Weak constraint presumes the opposite. An intermediate zone demarcates the belief that competence is widely shared, and that there is no single best way for government to proceed. The choice to constrain the substance of bureaucratic decisions proceeds from the idea that democratic government exists to serve popular ends. In contrast, the choice to control the procedures bureaucrats use to make decisions emanates from the idea that a democratic government is one that safeguards the liberties of its citizens.

The logic that links constraint with costs is one of probability. Tight constraint of bureaucratic behavior leaves little room for flexibility or for the exercise of bureaucratic expertise. It is thus more likely to incur costs to agency effectiveness. Inasmuch as factors beyond the bureaucrat's control are more likely to prevent the achievement of specified goals than they are to prevent the implementation of specified procedures, enforcement costs are more likely to arise when the substance of bureaucratic decisions is constrained.

The structure of choice among potential control mechanisms is constant, but the policy arenas in which the controller must make those choices vary considerably. Variations

both in the technology bureaucrats manipulate and the environment in which they operate mean that the same choices about values and costs may not be appropriate for all bureaucracies.

Differences among policy areas in technology and environment may lead controllers to different assessments of the relative capabilities of citizens, elected officials, and bureaucrats. In areas involving highly specialized expertise, for example, bureaucrats often have a greater advantage vis-à-vis citizens and elected officials than in areas where the technology is simpler and more accessible. Differences in technology and environment may also affect the importance controllers attribute to liberty and popular action as goals for a particular policy arena. For example, controllers may give greater importance to liberty in policy environments characterized by substantial goal conflict and more emphasis on accomplishing substantive goals in policy environments where there is considerable agreement on policy ends.

Technological and environmental variation can also affect the likelihood that control will create enforcement or effectiveness costs. For example, policy areas with highly uncertain technologies are more susceptible to the effectiveness problems associated with tight constraints because the uncertainty of the technology means that bureaucrats need considerable latitude to try various approaches to solving a problem. Where technology is certain, controllers are more likely to know how bureaucrats can achieve agency goals and can therefore specify detailed behavior without impairing the work of the agency.

Controllers are not alone in assessing the likely effects of control. Bureaucrats, too, have a vital interest in efforts to control them and potent resources to resist that control. Their stake lies in their political preferences, professional

values, and day-to-day work experience; at times it involves their very careers. The likelihood of bureaucrats resisting particular kinds of control also depends on the policy arena involved. Bureaucrats working with different technologies and in different environments tend to have quite different attitudes about their jobs. These attitudes, in turn, have consequences for the kinds of control bureaucrats resist and the kinds of control they accept. As we saw in chapter 6, for example, educational administrators see themselves as masters of a specialized technology; thus they consider themselves to be uniquely competent to make decisions about a broad range of issues. Their colleagues in fire and housing see their monopoly of expertise extending over a more limited domain. They are more likely to attribute competence to elected officials, and as a result are more willing to accept some external controls.

Bureaucratic resistance is certainly a force that makes control more costly to exercise, but it is not only that. Although some complaints are sheer self-interest, bureaucrats also express views about control that reflect an on-the-scene understanding of the consequences of control, and so call attention to other costs of control. In this way, bureaucratic attitudes illuminate not only what is possible but also what is desirable. Educators do possess substantive expertise, and their plaints remind us that tight constraints on the substance of educational decisions may indeed impose effectiveness costs on the work of an agency. Similarly, housing administrators insist on their need for flexibility because of the real technological demands of their work. Tight procedural constraints that deprive them of flexibility may seriously undermine their effectiveness.

Would-be controllers have difficult choices to make. Choosing a control mechanism ideally involves understanding

the means available, the values at stake, and the probable costs. It also involves calibrating decisions to specific policy arenas. Controllers who approach this task ignoring such differentiation lose the benefit of the richness a differentiated understanding of control provides. At times they may achieve their goals in any case, but those who pursue a single vision of control often limit their own ability to achieve those goals.

The Context of Control

No form of democratic control of bureaucracy exists in a vacuum; all are embedded in a broader political context. Control is exercised by citizens, legislators, or elected executives who play many roles besides controlling bureaucrats. Since these roles are not independent of one another, controlling bureaucrats cannot be looked at in isolation from other aspects of the political system. This embeddedness in the political context is important in at least two ways. First, control brings with it both costs and benefits that ripple through the political system. Second, the effectiveness of individual control mechanisms is often affected by conditions elsewhere in the political system. Our ability to control bureaucracies is limited by the basic attributes of our political system. Thus, a second lesson of this book for would-be controllers is that they must be sensitive to the context in which control takes place.

Controlling bureaucracies can create considerable opportunity costs for controllers. It requires time and effort both to formulate constraints and then to monitor bureaucrats to see that they are followed. This time and effort could be directed to other endeavors. One of the reasons citizens delegate power to representatives and

representatives delegate power to bureaucrats is to avoid having to make all the decisions.^[1] The more decisions bureaucrats are free to make on their own, the fewer citizens or their representatives must make. Conversely, the more tightly bureaucratic behavior is constrained, the greater the demands placed on citizens or their representatives. Since affecting governmental decision making is a peripheral concern for most citizens, and since controlling bureaucracies is only one of many activities public officials pursue, the opportunity costs of tight constraint can be significant. A president who spends a great deal of time supervising the details of the day-to-day operation of an agency is a president who has less time to spend ensuring the passage of a legislative program.^[2] Citizens who choose to vote on who will be awarded a city's recycling contract instead of having the decision made by bureaucrats are citizens who have less time to consider whom to elect to public office, as well as less time to devote to activities outside politics.

On the other hand, controlling bureaucracies can also bring benefits beyond the exercise of constraint. When citizens sit on review boards that oversee an agency's work, for example, they are controlling a bureaucracy; but they are also participating in governing themselves. To some theorists, the act of participating is more valuable than the control of bureaucracy it brings. They believe that participation itself achieves such desirable ends

[1] Robert A. Dahl calls this the criterion of economy and discusses it in After the Revolution ? (New Haven: Yale University Press, 1974).

[2] Herbert Kaufman, writing about intraorganizational control, makes an analogous point. He argues the attempt to monitor subordinate behavior closely "obliges executives to divert their attention and energies from strategic problems of policy and of external relations to details of internal operations. In short, it absorbs a higher proportion of organizational resources" (Kaufman, Administrative Feedback [Washington DC: Brookings Institution, 1973], p. 54).

as the creation of a healthy and fulfilled citizenry.^[3] For them the control-related costs of participatory institutions may fade in importance compared to the broader benefits they create. Similarly, when client groups play a role in agency affairs, they are contributing to the control of bureaucracy and also gaining recognition as significant groups in society. Those who place great value on the latter benefit may choose control strategies more for their consequences for the groups involved than for their impact on bureaucratic power. External control may also serve other functions within the bureaucratic context itself. Citizen participation, for example, may legitimate bureaucratic activities even if these activities are merely explained to the participants and not changed by them.^[4]

By the same token, however, other aspects of the political context enter into the success of mechanisms for controlling bureaucracy. When ordinary citizens are asked to participate in overseeing bureaucracies, the same factors that discourage them from political participation in general are likely to discourage them from participation in the control of bureaucracy. When interest groups are integrated into the processes of agency governance, the political

[3] Rousseau was probably the foremost proponent of this view. As Carole Pateman explains: "Rousseau's entire political theory hinges on the individual participation of each citizen in political decision making and in his theory participation is very much more than a protective adjunct to a set of institutional arrangements; it also has a psychological effect on the participants, ensuring that there is a continuing interrelationship between the working of institutions and the psychological qualities and attitudes of individuals interacting with them" (Pateman, Participation and Democratic Theory [Cambridge: Cambridge University Press, 1970], p. 22).

[4] In Bureaucratic Justice (New Haven: Yale University Press, 1983), p. 140, Jerry L. Mashaw suggests that this may be one benefit of participation in disability decision making. Daniel A. Mazmanian and Jeanne Nienaber make a similar point about the Army Corps of Engineers in Can Organizations Change? Environmental Protection, Citizen Participation and the Corps of Engineers (Washington DC: Brookings Institution, 1979).

forces that encourage only some groups to organize will affect the way democratic control of bureaucracy is achieved. When members of Congress control bureaucrats, fragmentation in Congress and the electoral incentives members face will affect the way they do so.^[5]

As we saw in chapter 2, political reality can clash with normative beliefs, and when that clash occurs, control mechanisms may malfunction. Control strategies that involve strong substantive constraint by legislators, for example, normally presume that legislators will act to secure general interests, or, alternatively, will act to resolve goal conflicts that might produce conflicting constraints. If legislators' incentives impel them to pursue particularistic interests, however, close legislative control of bureaucrats cannot achieve the values expected from it. Furthermore, such behavior may reinforce the bureaucrats' perception that politicians pursue only narrow self-interest. This entrenches the bureaucrats' conviction that control by elected officials should be resisted. Jerry Mashaw contends that legislative control, in fact, makes many of the problems with the bureaucracy in the Social Security administration worse, not better. He argues: "If bureaucracy in the pejorative sense of insular, rigid, and insensitive is the problem, then, as we have seen, reform in the image of democratic ideals is not the solution for a program like the DI [disability insurance] program. It is responsiveness

[5] Douglas T. Yates argues that congressional control of bureaucracy is diluted by the fragmentation of the committee system (Yates, Bureaucratic Democracy [Cambridge MA: Harvard University Press, 1982], p. 159). More generally, James Q. Wilson maintains that "the particularistic and localistic nature of American democracy has created a particularistic and client-serving administration. If our bureaucracy often serves special interests and is subject to no central direction, it is because our legislature often serves special interests and is subject to no central leadership" (Wilson, "The Rise of the Bureaucratic State," in Nathan Glazer and Irving Kristol, eds., The American Commonwealth, 1976 [New York: Basic Books, 1976], p. 103).

to the democratically constituted legislature that has pushed SSA in 'bureaucratic' directions."^[6]

Legislatures, of course, are not the only context in which control takes place. Client-oriented control strategies generally presume that all interests will organize. If, however, some groups in society do not have the political resources to organize, those groups will not be represented in the interplay of interests such strategies presume. Judicial review of bureaucratic decisions is equally limited by the scope of judicial authority and the powers of the courts to enforce their decisions. Moreover, any form of control will be afflicted by contradictory impulses if the political system itself is in conflict over what an agency should do.

In each of these cases, malfunctions in mechanisms designed to control bureaucrats are not the result of narrow flaws in design, but are manifestations of broader political phenomena. At times even an advocate of democratic control might applaud bureaucratic resistance to constraints that seem to serve the private interests of the controller. Such applause is generally rooted in ambivalence about how our political system is working. Championing appointed officials over elected ones is hardly the democratic norm. A dislike for the ways legislators, executives, or citizens seek to control bureaucracies suggests deeper problems with the way those actors wield political power, problems that have ramifications well beyond the issue of controlling bureaucracies. Democratic control of bureaucracy is hardly immune to the forces shaping other aspects of our political system. Often the road to improving control is blocked by the thorny democratic context in which control takes place.

[6] Mashaw, Bureaucratic Justice, p. 223.

A Strategic Approach to Control

A control relationship is in essence one of inequality. For control to be effective, no matter how it is exercised, the controller must have sufficient resources to induce the object of control to behave in the way the controller wishes. Control becomes difficult when the inequality runs in the opposite direction: when the object of control has greater resources than the would-be controller. In the case of democratic control of bureaucracies, the dominant resource used to exercise control is the controller's formal authority. Elected officials have the legal power to set up mechanisms that constrain the behavior of appointed ones, and they have a monopoly over that power. However, bureaucrats have other resources in considerably greater measure than their would-be controllers. These inequalities serve to hinder, not facilitate control.^[7]

Three critical resources possessed in abundance by bureaucrats put them in a favorable position compared to their controllers. First, they have responsibility for making the day-to-day decisions in their agency. Second, because they are on the scene, bureaucrats, more than anyone else, have information about the affairs of their agency. Third, bureaucrats command the resource of substantive expertise: they bring to their work specialized knowledge about how to achieve the ends of the agency.

None of these inequalities is either accidental or perverse; they flow from the very nature of bureaucracy. We

[7] In a more general discussion of the problem of control, Robert A. Dahl argues that one of the reasons representatives may not be able to control subsystems is the access of subsystems to resources that impede the exercise of control (Dahl, Dilemmas of Pluralist Democracy [New Haven: Yale University Press, 1982], pp. 50–51). Janet A. Weiss and Judith E. Gruber also discuss the point in "Using Knowledge for Control in Fragmented Policy Arenas," Journal of Policy Analysis and Management 3 (1984): 225–47.

hire bureaucrats because we cannot deliver services alone. Bureaucrats are hired precisely because they are specialists, and they are deliberately given control over agency affairs. We need both their substantive expertise and their undivided attention to conduct the business of government. Alas, these very virtues become liabilities when it comes to exercising democratic control. They create the conditions that give rise to the costs of control.

Bureaucrats are able deliberately to ignore a constraint or to let it slip in the course of the exigencies of agency life because they have the resource of responsibility for day-to-day decision making. When they ignore democratic control they produce the cost of bureaucratic resistance. When they let it slide, they create circumstances that give rise to the problem of enforcing constraint. Having day-to-day responsibility also means that bureaucrats are uniquely qualified to calibrate their actions to changing conditions; tight constraint denies bureaucrats the ability to be as effective as possible. Bureaucratic dominance of all three resources means that bureaucrats are in a privileged position to know what is going on in their agency and how to achieve agency ends. When controllers have limited access to these resources, they are unable to determine accurately what kinds of constraints on bureaucratic behavior will produce the ends they desire. It is this situation that produces the effectiveness cost created by the imposition of the wrong constraint.

Democratic norms ignore the difficulties with such problems of control because they stress the sovereignty of the citizen and the subservience of the bureaucrat. But authority to command bureaucrats guarantees neither that the command will be obeyed nor that control will achieve what the controllers want it to. Because of the abundance of other resources bureaucrats possess, formal

authority is not enough to secure control of bureaucracies.^[8] A third lesson of this research is thus that controllers must look at control as a strategic problem of creating and marshaling resources to facilitate the exercise of control.

One strategy for improving the prospects for control is to use formal authority to force bureaucrats to share their resources, especially information, with would-be controllers. Michael Lipsky argues for the need "to provide a better balance of power between street-level workers and clients". He suggests that one way to do this is to "demystify" bureaucratic processes by providing clients with information such as guides to client rights, maps of the bureaucratic system, and written summaries of agency actions.^[9] Control mechanisms themselves may be designed and used to narrow the resource gap between controllers and bureaucrats by opening up an agency to public examination. "Sunshine" laws that require exposing some bureaucratic processes to public view, for example, may be used not merely as a procedural constraint but also as a way of increasing the amount of information controllers have about agency affairs. Such increased information may in turn allow for tighter substantive constraints without incurring effectiveness costs, since well-informed controllers are less likely to impose the wrong constraints. Similarly, legislatively mandated reporting requirements may serve not merely as a way of overseeing what bureaucrats have done but also as a way of providing information about bureaucratic activities to citizens and

[8] This perspective on authority is similar to that proposed by Chester I. Barnard. He argues in effect that authority lies in the eye of the beholder and not in the person of the possessor. As a result he maintains that people cannot merely be ordered to cooperate but must be induced to (Barnard, The Functions of the Executive [Cambridge MA: Harvard University Press, 1983], chap. 12).

[9] Michael Lipsky, Street-Level Bureaucracy (New York: Russell Sage Foundation, 1980), p. 195.

interest groups. Armed with the bureaucrats' own information, they may exercise control more effectively.

Would-be controllers can also improve their prospects for control by extending their sights to resources other than formal authority as resources for control. As we saw in chapter 4, the value of democratic control pales in bureaucratic eyes in the light of the demands of their jobs. Thus, a claim by controllers to democratic authority is not a potent resource if bureaucrats see control as threatening other values. But, as we saw in chapter 6, these other values, and particularly the bureaucrat's desire to do a good job, provide a vehicle for the introduction of control. Successful controls are often those that can ride piggyback on the bureaucrats' conception of their self-interest. As William Niskanen argues, "Any public administration reforms that will be both beneficial and enduring must create conditions such that the activities (legal or otherwise) of bureaucrats and review officers, acting in their personal interests, are more nearly consistent with the public interest."^[10] One way of making the private interests of bureaucrats converge with the public's interest in control is more explicitly to parlay the resources controllers command and bureaucrats seek into bureaucratic acceptance of control.

Controllers do possess an array of resources potentially useful to bureaucrats. Funding, access to political leaders, information, and cooperation are all things bureaucrats value. All are resources that can ease the process of control. Political executives with close connections to the White House, state house, or city hall may shore up their

[10] William A. Niskanen, Jr., Bureaucracy and Representative Government (Chicago: Aldine-Atherton, 1971), p. 194. Peter M. Blau similarly argues that public control of bureaucracy hinges upon "converting external dysfunctions into internal needs of the organization that disturb its personnel" (Blau, The Dynamics of Bureaucracy [Chicago: University of Chicago Press, 1963], p. 264).

influence within their agency by using those connections to deliver funding, jurisdiction, or other "goods" bureaucrats want.^[11] Legislators may gain control of some bureaucratic decisions by being loyal supporters of an agency's budget.^[12] Mayors and governors can use discretionary funds to induce bureaucrats to pursue programs they otherwise might not.

Citizens have the resource of cooperation and the threat of disrupting bureaucratic activities if they withhold cooperation.^[13] Most bureaucrats would prefer not to have their agency be the object of boycotts and demonstrations, and they may accede to some of the wishes of citizens in order to avoid such pressures. On the positive side, public cooperation can be helpful to bureaucrats. Parents know about their children's needs and have some influence over their children's behavior in school. Tenants can cooperate in maintaining a housing project. Citizens can report suspicious activities to fire officials. All of these resources are valued by bureaucrats, and may thus be converted into a measure of control. Moreover, the strategy of breaking the bureaucratic monopoly on information may augment the effectiveness of these resources. Information about bureaucratic activities may mobilize previously apathetic citizens by informing them of the consequences of bureaucratic action, thereby encouraging them to harness their resources to the task of control.^[14]

[11] "In every department in every recent administration, one of the chief ways political executives gained support in the bureaucracy was by being, or at least appearing to be, their agency's vigorous spokesman," writes Hugh Heclo (A Government of Strangers [Washington DC: Brookings Institution, 1977], p. 196).

[12] Douglas Arnold, Congress and the Bureaucracy (New Haven: Yale University Press, 1979).

[13] Lipsky argues that clients have resources to "impose a variety of low-level costs" on bureaucrats (Street-Level Bureaucracy , p. 57).

[14] David L. Kirp, for example, notes that a mid-1960s audit of Oakland's Title I expenditures provoked anger, and then mobilization in the black community

when it showed that funds had not been targeted at disadvantaged children (Kirp, "Race, Schooling and Interest Politics: The Oakland Story," School Review 87, no. 4 [August 1979]: 355–97). More generally Mashaw argues that "a sense of illegitimacy, even outrage, may also have its value in promoting bureaucratic justice" (Bureaucratic Justice , p. 142).

A further strategy for improving the prospects for control is for would-be controllers to create arenas in which their resources can be utilized. If resources are to be transformed into control, bureaucrats must know that would-be controllers have these resources and controllers must have the opportunity to use them.^[15] Given the normal proclivity of bureaucrats to keep to themselves, such opportunities are not always easy to find. To promote these opportunities, control structures can be designed to force greater interaction between bureaucrats and citizens or elected officials and in that way facilitate the conversion of resources into the acceptance of control.

Control mechanisms such as advisory boards or consultative committees that provide for such interaction achieve control of bureaucracy in two ways. First, control emerges from the explicit work of the board or committee itself. Second, by creating arenas for interaction, such control mechanisms provide opportunities for informal influence that may go beyond the formal arrangements. For example, school councils are important not only because of their official powers but also because they provide a forum for bringing parents and school administrators together. Such interaction provides parents with an opportunity to proffer their resources of information and cooperation and to express views on how they would like the schools to be run. Because the resources are useful to administrators, they may be willing to listen.

[15] In The Logic of Bureaucratic Conduct (Cambridge: Cambridge University Press, 1982), Albert Breton and Ronald Wintrobe argue for the role of interaction as a vehicle for creating the trust that they assert is essential for exchange to take place.

Listening may in turn result in changed behavior. Having listened, administrators may have new sensitivity to certain problems, enhanced appreciation of parent competence, or simply the desire to keep the parents coming back. Furthermore, to the extent that bureaucrats are controlled by a process of anticipating what citizens will accept and trying to stay within those bounds, listening should improve control by clarifying just where those bounds really are.^[16] Mashaw argues that client participation in disability determinations "may help to ensure that the adjudicator 'really listens' to what the claimant thinks is important and that disadvantageous evidence is not accepted without close scrutiny."^[17] Institutions such as ombudsmen may also provide an arena for interaction to take place. Thus they can control bureaucrats not merely through their official power to redress grievances but also by providing bureaucrats with valued information about citizen unhappiness.

Arenas for the exchange of resources may also be created through modest administrative restructuring to reduce the insularity of bureaucratic life.^[18] As the data in chapter 4 indicate, bureaucrats rarely seek out contact with other public officials or with citizens; the contact they have generally comes from structured interactions such as board meetings, the need to deal with other officials to

[16] Anthony Downs argues that pressure on bureaucrats to improve their performance "tends to make each official shift his definition of 'satisfactory performance' closer to the definition held by these external agents. This is the main way in which social agents get a bureau to perform its functions to their satisfaction" (Downs, Inside Bureaucracy [Boston: Little, Brown, 1967], p. 193).

[17] Mashaw, Bureaucratic Justice , p. 140.

[18] William R. Dill argues that different organizational environments produce varying levels of informal interaction among organizational members, and that such interaction in turn reduces autonomy (Dill, "Environment as an Influence on Managerial Autonomy", Administrative Science Quarterly 2 [March 1958]: 409–43). Mashaw also argues that bureaucracies can be structured "to establish appropriate relationships" among political actors (Mashaw, Bureaucratic Justice , p. 158).

secure supplementary funds, or citizen-initiated requests. Since bureaucrats have little time or inclination to consult with others over broad policy issues, they may resist control mechanisms directed to this end. If, however, more contact can be built into daily coping processes, controllers gain an opportunity both to demonstrate their competence to bureaucrats and to offer useful resources. In this way they may gain greater control over bureaucratic behavior. Lipsky, for example, points out that street-level bureaucrats are at times compelled to make their decisions in public "so that they can be exposed to, if not influenced by, the presence of those affected by the decisions. Such policy incorporates the theory that clients are likely to become more a part of the bureaucrats' reference groups if they are present at times when decisions are made."^[19] Mashaw makes a similar proposal for structuring "the state agency process so that claimants are as real to adjudicators as are the medical consultants, unit supervisors, bureau chiefs, QA [quality assurance] staff, medical listings, DOT [Dictionary of Occupational Titles ], DISM [Disability Insurance State Manual], and disability determination forms that make up their daily work environment. The examiners could be forced to talk to claimants, to treat them as important sources of information, to explain their eligibility decisions."^[20]

Cross-national research offers strong evidence of the impact of institutional arrangements on bureaucratic autonomy. In a study of seven western democracies, Aberbach, Putnam, and Rockman examine the attitudes of bureaucrats and politicians and the relationships between the two groups. Although they find considerable similarity in the attitudes held by each group across countries,

[19] Lipsky, Street-Level Bureuacracy , p. 128.

[20] Mashaw, Bureaucratic Justice , p. 198.

they find noticeable differences in the nature of the linkages in countries with decentralized political systems compared to those with more centralized governments. They conclude: "Bureaucrats must do some things and may do other things. Attitudes affect what individuals choose to do among the repertoire of those things they may do, but what must be done is greatly influenced by institutional arrangements. Institutional arrangements that fragment power create conditions of mutual dependence, and mutual dependence in turn encourages interaction between politicians and bureaucrats, because each holds resources valued by the other."^[21] Fragmentation does not always hold the answer to bureaucratic insularity. Given enough dispersion of power, bureaucrats can shop for the resources they need without relinquishing control. Under those circumstances concentrating resources in the hands of mayors, councils, or governors may provide the structural lever to force bureaucrats to depend more on elected officials.

Control as Exchange

By emphasizing the creation of resources and the development of arenas for their strategic use, I have moved away from traditional conceptions of control as based in an authority relationship toward an understanding of control based more on a model of exchange.^[22] In an exchange

[21] Joel D. Aberbach, Robert D. Putnam, and Bert A. Rockman, Bureaucrats and Politicians in Western Democracies (Cambridge MA: Harvard University Press, 1981), p. 236.

[22] Charles E. Lindblom discusses three dominant modes of control: exchange, authority, and persuasion. While exchange is generally associated with economic arrangements, we may borrow from Lindblom and consider how this mode of control might be incorporated into strategies for democratic control of bureaucracy (Lindblom, Politics and Markets [New York: Basic Books, 1977]).

change model, control results not from political actors telling bureaucrats what to do but from constructing conditions in which bureaucratic behavior is constrained in exchange for resources that bureaucrats seek.^[23] Such control emanates from a process of interaction, not from one of orders from above.

Control by exchange, of course, operates against a backdrop of authority. Formal governmental powers provide many of the resources, such as public funds and legal jurisdiction, that are used in exchange. Authority may also provide the threat of worse consequences if exchange is not pursued. Public officials have the authority to strip recalcitrant administrators of their funding and, in some cases, of their jobs. Citizens have the authority to bring a lawsuit against an agency that is uncooperative. Administrators understand this and may often be willing to engage in exchange to avoid such consequences.^[24] The two modes of control thus work closely together.

Sensitivity to the prospects for exchange should improve

[23] Students of both private and public organization have examined exchange as a means for both intraorganizational and interorganizational control. David Jacobs, for example, argues that organizations are dependent on those actors who control items essential to the organization that cannot easily be obtained from other sources. He suggests that organizations will mold their behavior to meet the expectations of those who control such resources (Jacobs, "Dependency and Vulnerability: An Exchange Approach to the Control of Organizations," Administrative Science Quarterly 19 [March 1974]: 45–59). Jeffrey Pfeffer and Gerald R. Salancik provide an extended discussion of the role of resource dependence in the control of organizations in The External Control of Organizations (New York: Harper and Row, 1978). Albert Breton and Ronald Wintrobe base their entire analysis of bureaucratic conduct on the assumption that bureaucratic relationships are exchange relationships (The Logic of Bureaucratic Conduct [Cambridge: Cambridge University Press, 1982].

[24] Aberbach, Putnam, and Rockman argue that "constitutionally politicians are everywhere empowered to reject the counsel of bureaucrats, although such rejection is infrequent in practice. Policy making is thus a kind of dialectic, in which the 'law of Anticipated reactions' normally governs the behavior of bureaucrats" (Bureaucrats and Politicians , p. 248). Peter H. Schuck discusses the use of tort law to control bureaucrats in Suing Government (New Haven: Yale University Press, 1983).

the design of formal control structures. Thinking in terms of exchange suggests new ways of conceptualizing strategies for democratic control. Orders are augmented or replaced by resources, commands by negotiation. Many of the same control mechanisms may be used, but used in a way that provides resources for exchange or facilitates participation in an exchange process.

The potency of exchange in a bureaucratic context can be seen in a variety of situations. Students of the federal government often decry the insularity of the "iron triangles" linking agencies, congressional subcommittees, and interest groups.^[25] Looked at from the perspective developed here, these iron triangles are testimony to the power of exchange through which interest groups and members of Congress are able to affect bureaucratic behavior by providing the agency with needed support. Similarly, in the area of coordination of human service delivery, Janet Weiss has argued for the importance of providing bureaucrats with incentives to change their behavior.^[26]

Acceptance and use of exchange are not limited to actors outside the bureaucracy. Bureaucrats themselves often use exchange to supplement their formal authority. "Street-level" bureaucrats need to gain client compliance, and Lipsky argues that they do so "either through the control of resources that the client desires (utilitarian compliance) or … through force or the threat of force."^[27] Good policemen mix persuasion and help with coercion in the course of their work.^[28] Many teachers overcome the

[25] See, for example, Francis E. Rourke, Bureaucracy, Politics and Public Policy (Boston: Little, Brown, 1969).

[26] Janet A. Weiss, "Substance vs. Symbol in Administrative Reform: The Case of Human Services Coordination," Policy Analysis 7 (Winter 1981): 21–45.

[27] Lipsky, Street-Level Bureaucracy , p. 42.

[28] See William K. Muir, Police: Streetcorner Politicians (Chicago: University of Chicago Press, 1977).

inadequacy of their authority by exchanging the relaxation of some classroom rules for order and concentration by students.^[29] In the regulatory arena Eugene Bardach and Robert Kagan observe that "an enforcement official's ability to win cooperation is rooted in the relationship of reciprocity or exchange that he manages to establish."^[30] Regulatory inspectors trade responsiveness, forebearance, and information for responsible social behavior on the part of the regulated.^[31]

Designing control mechanisms to coincide with bureaucrats' perceptions of their own interests may appear to violate the very premise upon which the need for control is based—i.e., that public actions should be controlled by the citizenry. To an extent this appearance is accurate. Strategic design allows for considerable bureaucratic power in decision making by accepting the facts that bureaucrats have resources for resistance that cannot merely be ordered away and have resources that often make them more able to discern appropriate courses of action than controllers. However, short of a fundamental change in the division of labor in democratic societies that eliminates the very existence of delegated authority, bureaucrats will have these resources and will be in a position to resist controls they do not find acceptable. Ignoring that fact may produce controls that are normatively

[29] Mary Haywood Metz, Classrooms and Corridors (Berkeley and Los Angeles: University of California Press, 1978), especially pp. 97–116.

[30] Eugene Bardach and Robert A. Kagan, Going by the Book: The Problem of Regulatory Unreasonableness (Philadelphia: Temple University Press, 1982), p. 130.

[31] Bardach and Kagan also contrast what they call indirect regulation with command and control regulation. The latter relies on formal legal authority, whereas many forms of indirect regulation, such as providing citizens with information to be their own inspectors and using liability law, bear a close resemblance to control by exchange (Going by the Book , pt. 3).

pleasing but ineffective in practice.^[32] Accepting that fact may allow controllers to develop strategies that widen the domain in which bureaucrats are willing to accede to controllers' wishes.^[33]

As every chapter in this book has indicated, democratic control of bureaucracy involves complex and difficult choices. These choices are made more difficult by the fact that bureaucrats command resources that controllers need. Fortunately, controllers also have resources that bureaucrats find useful. The road to successful control lies in understanding both the democratic and bureaucratic sides of the problem and carefully crafting efforts so that neither democratic norms nor bureaucratic facts are denied.

[32] Heclo argues for the virtues of evoking "conditional cooperation" as opposed to "invoking authority" (Government of Strangers , p. 220. Similarly, Martin Landau and Russell Stout, Jr., call for seeking decisions that fall within the "zone of acceptance" of subordinates (Landau and Stout, "To Manage is Not to Control: Or the Folly of Type Two Errors," Public Administration Review 39 [1979]: 151).

[33] Barnard calls this the "zone of indifference" (Functions of the Executive , pp. 167–70).