Resumé

This article attempts a systematic comparison of how policy is implemented in Sweden and the United States. We suggest that differences might be explained by examining policymaking processes, policy content, and governmental structure. The comparison reveals differences in "incentive structure'. Both Sweden and the U.S. have responded to the complexity and uncertainty that accompanies increased on central government by adopting "softer' mechanisms for implementation. The effect of these shifts in both systems is to devolve important decisions to lower levels of government. In Sweden, this devolution is still accompanied by strong incentives for national policymakers to undertake implementation, although the incentives may not be as strong as they were in the past. In the U.S.. it is accompanied by a radical weakening of incentives to carry out implementation and increasing in policy content and relations among key actors.

Introduction: Comparing Implementation Processes

In this paper an attempt is made to engage in systematic comparison of how policy is implemented in Sweden and the United States. The purpose is threefold: (1) to state explicitly how implementation processes differ in the two countries: (2) to suggest how differences in implementation might be explained by examining policymaking processes, policy content, and governmental structure; and (3) to suggest how changes in these factors are likely to affect implementation. We have tried to keep our conceptual framework as simple as possible, in an effort to specify what we regard as the major questions underlying comparisons of policy implementation among countries.

Our model suggests that policymakers and administrators work within an incentive structure (Fig. 1). This structure is defined by formal and informal constitutionalconventions, norms, and political organization, all of which can change, but are likely to be relatively stable overtime. Policies are responses by politicians and administrators to political demands, worked out in the context of this incentive structure. Policies carry at least three types of messages. The first is the administrative responsibilities that are expected to follow from policy:

Which agencies and officials are responsible for what actions? The second is the relationship among levels of government that is expected to follow from policy: Which level of government is responsible for what actions? The third is the mechanisms or 'implements' that are to be used to express these responsibilities and relationships: What rules, procedures, agreements, and organizational structures follow from the administrative responsibilities and relations among levels of government laid out in the policy? To the degree that governments differ systematically in the incentives under which policymakers and administrators work, they can be expected to differ systematically in their choice of implements. To the degree that they differ systematically in their choice of implements, they can be expected to differ systematically in the way policies are implemented. From these differences, we can learn alternative ways of implementing public policy. Isolating and explaining these systematic differences is the task we have set for ourselves in this paper. Our analysis falls far short of exhausting the possibilitiesfor and it is more concerned with specifying dimensions of variation than with testing them empirically.

Sweden, at first glance, fits the simple stereotype of a culturally homogeneous, geographically compact, hierarchically ordered governmental system. From this stereotype, we might infer that the implementation of national policy is less complex,less less uncertain, and therefore more likely to succeed

than in the U.S. The U.S., on the other hand, is from outward appearances a country of great cultural diversity, enormous geographical scale and variety, and a federated governmental structure, from which we might infer that implementationof policy would be relatively complex, conflict-ridden, highly uncertain, and therefore less likely to succeed than in Sweden.

There are several problems with these simple stereotypes. First, the explanatory variables are not well-specified. Cultural homogeneity, geographical compactness, hierarchical structure are relatively difficult to define in ways that have a direct relationship to policy. Second, it is not clear that homogeneity, compactness, hierarchy have a simple, linear relationship to complexity, conflict, and uncertainty in the implementation of national policy. The perception that it is 'easier' to implement national policy in Sweden than in the U.S. may simply be wrong, or the two systems may raise different problems that make it equally difficult implement national policy. Third, differences between two political and governmental systems are not constant over time. Both Sweden and the U.S. have undergone significant changes in relations between national and subnational These changes reflect shifts in thinking both about the role of national government and the process by which national policy is to be translated into practice. In other words, the simple stereotypes may misrepresent the way experience with policy implementation has changed relations among levels of government in the two countries over time.

Our objective in this analysis is to move beyond these stereotypes and to base the comparison of policy implementation on factors more directly related to the content of policy and the structure of government. Our method is, first, to specify the incentive structure under which policymakers work; second, to specify how these incentives affect the content of policy and the likelihood that policymakers will attend to implementation issues; third, to specify how the content of policy and the structure of relations among levels of government interact; and fourth, to specify how changes in political incentives may affect implementation.

In brief, our comparison of Sweden and the U.S. reveals the following conclusions:

1. Policymakers have stronger incentives to attend to implementation issues in
Sweden than in the U.S.

2. This difference in incentive structures is reflected in the content of policy. National policy in Sweden relies more heavily on the implements of consultation, and graduaJ evolution of policy requirements into governmental structures. National policy in the U.S. relies more heavily on the implements of distribution formulas, procedures, assignment of rights and responsibilities, and regulations.

3. The interaction of policy content with governmental structure in Sweden produces relative stability in relations among key actors at different levels of government and a relatively greater reliance on informal negotiation to resolve implementation problems. In the U.S., the interaction of policy content and governmental structure produces relative instability in relations among key actors at different levels of government and a relatively greater reliance on changes in policy, formal rules, and procedures to resolve implementation problems.

4. Both Sweden and the U.S. have responded to the complexity and uncertainty accompanies increased demands on central government by adopting 'softer' implements, policies, and implementation strategies that entail increased on lower levels of government to implement national policy. But the consequences of these changes are very different in the two systems. In Sweden, accommodation to complexity and uncertainty occurs by informally adjusting implements within the context of broadly defined policy objectives. In the U.S., the accommodation occurs by constantly changing implements and objectives.

5. Both Sweden and the U.S. are undergoing significant shifts in the political incentives that operate on national policymakers. The effect of these shifts in both systems is to devolve important decisions to lower levels of government. In Sweden, this devolution is still accompanied by strong incentives for national policymakers to attend to implementation, although the incentives may not be as strong as they were in the past. In the U.S., this devolution is accompanied by a radical weakening of incentives to attend to implementation increasing instability in policy content and relations among key actors.

As in all comparative analyses, these conclusions are matters of degree rather than absolute differences. They require development, qualification, and careful empirical testing. The analysis proceeds in the following sections. The first addresses differences in political incentives and their effect on policy content. In the next the analysis concerns the interaction of policy content and governmental structure in the two systems. The final section deals with changes in political incentives and their effect on implementation.

Political Incentives and Policy Content

The connection between political incentives, policy content, and implementation can be described in at least three ways. First, the incentive structure within which policymakers work may encourage or require varying degrees of consultation with people who play an important role in implementation. Policymakers in

different incentive structures will ireflect different degrees of concern for implementation problems. Second, policymakers may, by virtue of the incentives under which they work, choose certaiin implements more often than others. The choice of implements influences the way implementation will proceed. And third, pattems of consultation in policymaking may carry over into the implementation process, influencing the way policy is adjusted to complexity, conflict, and uncertainty.Solutions implementation problems may be more likely to arise in some incentive structures than in others.

The U.S.: Instability and Institutional Distrust

National policymaking in the U.S. is characterized by (1) relatively weak incentives for elected officials and political appointees to be interested in policy implementation, and very strong incentives to be interested in major policy changes; (2) broad consultation with affected interest groups, but relatively little reliance on the advice of implementing bureaucracies; and (3) unstable, episodic changes in the role specified for state and local government.

Frequent elections, instability in electoral politics, and a weak party structure have profound effects on the content of national policy in the U.S. Legislators draw relatively little electoral support from national parties or national candidates. Their re-election depends largely on their ability to give the appearance of delivering benefits directly to constituents. This incentive structure rewards individual entrepreneurship, rather than party loyalty, and forces legislators to calculate the benefits of their actions by the amount of credit they can accumulate with their electoral constituencies in regular two-, four-, and six_-year cycles (Mayhew 1974). Their work on legislative committees, their advocacy with governmental agencies on behalf of constituents, and their visits home are calculated demonstrate the direct benefits of legislative action for their electoral districts. The bulk of their time is spent on activities directly related to re-election, rather than to broader, more complex questions of policy (Fenno 1973, 1978). Concern for implementation of past decisions and oversight of administrative actions assume a relatively low priority (Rosenthai 1981, Ogul 1981). As legislators more stable electoral constituencies, through re-election, these activities become potentially more politically rewarding. But implementation and administrative oversight still compete with other activities that carry greater political rewards - leadership within legislative bodies, seeking higher office, and participation in networks of influence. In this incentive structure, the long-run effects of policy are less important to legislators than the short-run tangible benefits distributed to narrowly-defined constituencies (Sundquist 1981).

Similar incentives operate on the U.S. executive branch. The incentives are embedded in the U.S. Constitution. The formal responsibilities of the executive and legislative branches are constructed to encourage each branch to assert its. independent authority in policymaking, rather than defer to the authority of the:

other or to form durable coalitions. The framers of the Constitution accomplished a separation of powers among branches by giving each a separate basis of authority and then requiring them to collaborate in the making of policy, 'giving to those who administer each (branch) the necessary constitutional means and personal motives to resist encroachments of the others'. By this device, 'ambition is made to counteract ambition' (The Federalist, Madison no. 51, no. 47, and no. 48). The resulting system is one based on a radical distrust of concentrated power and a belief in institutionalized distrust as a way of preventing it.

Since presidents have shown little interest in their role as chief executives, the executive role nominally devolves to cabinet-level department heads and their immediate staffs, who are political appointees. This executive elite operates in an incentive structure characterized by conflicting loyalties. In order to be influential as initiators of policy they must cultivate close ties both with congressional committees the Executive Office of the President. Too much attention to congressional damages their effectiveness within the administration; inflexible loyalty to the administration's program limits their ability to bargain effectively with influential legislators. To be effective managers of their agencies they must be loyal both to the President's program and to the interests of career civil servants. Too strong an advocacy of their agencies' interests undermines their influence within the administration; inflexibility on matters of policy within their agencies limits their influence with career staff. Furthermore, the number of political appointees (in excess of 3,000 by conservative estimates), the instability of their tenure (an average of less than eighteen months), and the great variety of roles they perform make it virtually impossible for a President to communicate a sense of cohesiveness on national policy (Allison 1980, Heclo 1977, 38-39). In the absence of strong direction from the center, political appointees tend to form alliances with interest groups, legislative staff, and career staff who share common substantive concerns. At the level of government where policy and administration intersect, relations among key actors tend to take the form of 'issue networks' - diffuse coalitions of mutual influence based on common expertise (Heclo 1978, 102-104).

The incentive structure of political appointees is much like that of legislators and the President. It emphasizes short-term successes, based on specific policy initiatives that deliver immediate tangible benefits to clearly-defined constituencies, than long-term plans with high administrative uncertainty and diffuse benefits.

Career civil servants are marginal actors in this structure. Presidents, legislators,and appointees are impatient for short-term changes. They view the permanent bureaucracy as preoccupied with stability and continuity, and thereforeunresponsive the political demands of elected and appointed officials. The development of policy, therefore, has increasingly devolved to specialized policy staffs reporting to elected or appointed officials in the Congress, the Executive

Office of the President and the administrative agencies, rather than to career officials. The dominant mode of thought among policy staffs is more like that of economists than public administrators; they focus on the benefits, costs, and political trade-offs among policy alternatives and manifest a relative disinterest in institutional problems entailed in implementing policy. The result is that questions of administrative feasibility are postponed for later consideration, or not considered at all in the formulation of policy (Hargrove 1983, Lynn & Whitman 1981).

Recent developments in relationships among national, state and local governmentin U.S. can be explained by the incentive structure in which national policy is made. The period of maximum growth in national policy, in the 1960's and 1970'5, was accompanied by dramatic increases in national expenditures, but only slight increases in the number of employees working for national government (Heclo 1978, 89-91, Elmore 1982). Major changes in domestic policy occurred during this period, but these changes were accompanied less by expansions of national government, per se, than by expansion in the use of money and rules by national government to influence state and local decisions. The first round of policy initiatives - Great Society programs in education, employment, housing and the like - consisted of mainly direct intervention from the national to the local level. These programs were explicitly redistributional in purpose, often based on adversarial assumptions about the relationship between national and local government. State and local government were assumed to be at odds with the redistributional priorities of national government; the role of national government was to use its resources to direct local attention to the needs of the disenfranchised and the economically disadvantaged. The second round of national policy initiatives - block grants and revenue sharing in housing, employment,social and the like - was characterized by a more conciliator/ approach. State and local governments were granted more discretion in the use of national funds, but the purpose of national policy was still assumed to be essentially redistributional. Under block grants and revenue sharing, states and localities became more explicit agents of national policy, setting up separate organizations within their jurisdictions to administer national laws and engaging in extensive evaluation and reporting to document uses of national funds. In the first period, national policy was addressed mainly to two constituencies - the intended beneficiariesof and the advocates of redistribution within state and local government, whose stature was increased by their connection with national programs. In the second period, national policy was addressed mainly to state and local governments as political constituents. National policymaking can be seen in terms of the relative influence of two sets of alliances - one, 'vertical' alliances based on the interests of administrators and individuals who were the beneficiaries of national policy, extending across levels of government; the other, 'horizontal' alliances based on common jurisdictional interests among units of

government at the same level (Beer 1978). In the first period, vertical alliances dominated; in the second, horizontal. In both periods, national policy was necessarily 'government by remote control', in which the national government used money and regulations to stimulate state and local attention to national objectives (Heclo 1978, 92), and 'third-party government', in which the actual delivery of services was carried on by a complex collection of public and private organizations operating in complex frameworks of government regulation and market forces (Salamon 1981).

The relationship between national, state and local government in the U.S. also finds its roots in the Constitution. Just as the interplay between the executive and legislative branches at the national level is designed to make ambition counteract ambition, so too is the federal system designed to put one level of government in a constant state of tension with another. 'Power being always the rival of power, the general [national] government will stand ready to check the usurpations of state governments, and [the states] will have the same disposition toward the general government'. The people, using their electoral power, will check the abuses of one level of government by 'using the other as the instrument of redress' (The Federalist, Hamilton, no. 28). Questions of jurisdiction among levels of government settled by the play of electoral politics - one level of government asserting claim to jurisdiction, another asserting a counterclaim. In all but a few policy areas, jurisdictional questions are 'settled' only temporarily by national policy; they are subject to continuous re-negotiation with shifting political constituencies.

Out of these political incentives, the national government has evolved a peculiar and distinctive style of policymaking in the past two decades. Policymakers their attention sequentially from individuals and social groups as the beneficiaries or targets of national policy to state and local government; the purpose of national policy vacillates between remedying inequities and externalities society, by taking an adversarial posture toward state and local government in their judgments about what policy ought to be. Which purpose dominates national policy at any given time is a function of whether vertical or horizontal alliances exercise the major influence over national policy. When vertical alliances or advocates of national intervention - exercise the major influence national policy focuses on the use of the federal government's financial and regulatory power to override and modify state and local preferences, often constructing nationally-financed -mandated program organizations within state and local government. When horizontal alliances - or advocates of state and local control - exercise the major influence, national policy focuses on devolving authority and discretion to states and localities, reducing the federal government's regulatory and financial role, and allowing state and local preferences to have a major effect on the content and structure of national policy.

Relatively short, fixed electoral cycles mean that perceptions of the dominant

political constituency can change rapidly, often every two-to-four years. The high value placed on delivering short-term, tangible benefits to well-defined constituenciesmeans the debates on the content of national policy focus on the narrow, distributional impact of national financing and regulation on key constituencies:private ethnic groups, cities, counties, states, school districts, etc. The adversarial nature of relations among levels of government means that national policy typically focuses on the formal division of benefits and responsibilitiesamong and has a relatively heavy regulatory content - spelling out Riles, procedures, rights, and remedies either in the law itself or in the administrative regulations that follow from it. Weak incentives for attention to implementation among elected and appointed officials, and the marginal role of career civil servants in policymaking, mean that the administrative problems precipitatedby policy are often not fully addressed before a new policy is introducedin place. Shifts in national policy are based less on evidence that a policy has 'failed' than on the perception that another constituency should be the primary beneficiary of national policy.

The effect of this incentive structure on policy implementation can be illustrated by examining, for example, national policy toward education in the U.S., which has undergone a series of reversals since the national government increased its presence in the field in 1965 (Elmore & McLaughlin 1982). Immediately after the passage of the Elementary and Secondary Education Act, the ambitious redistributive goals of the legislation seemed to be diluted by deference on the part of federal administrators to state and local education authorities. In 1969, however, a coalition of civil rights groups issued a report critical of the administration federal policy, and that report stimulated more specific legislative standards for the use of federal monies and tougher administration (Murphy 1973). Federal enforcement seemed to stabilize at a relatively high level during the 1970'5, when several unsuccessful attempts were made to introduce federal block grant legislation in education. Then, in the mid-1970'5, Congress passed its most stringent regulatory act in education - a law designed to provide detailed procedural and programmatic guarantees for handicapped children. This law has stimulated considerable resistance and outright opposition toward national intervention education, and resurgence of state and local education authorities as political influentials (Hargrove et al, 1981). Recently, the ideological climate, if not the details of policy, have shifted back to deferential treatment of states and localities.

Similar patterns of policymaking and implementation can be seen in the regulationof safety and health (Kelman 1980, 1981), civil rights enforcement (Rabkin 1980), employment, and housing (Williams 1980). The essential instability of policy, the lack of sustained attention to implementation questions, the ebb and flow of influence by vertical and horizontal alliances, the focus on funding formulas, standards, rights, and responsibilities as the implements

of policy are all characteristic of U.S. policy to one degree or another. These attributes are not coincidental. They are a direct result of a political incentive structure that rewards short-term results over long-term systemic changes, changes in policy over adjustments in implementation, shifts in political constituenciesover alliances, and institutionalizes distrust of centralized authority.

In this incentive structure, compliance with rules becomes more important than the long-term effectiveness of the policy. It is possible to demonstrate compliance in the short-term, but not significant changes in the behavior of private firms or state and local government. Furthermore, the political tide can turn against a policy very quickly, reversing earlier gains in implementation. Those who wish to demonstrate effects must concentrate on the narrowest indicators or successes.

Sweden: Stability and Distrust of Conflict

National policymaking in Sweden is characterized by (1) relatively strong political incentives for policymakers to attend to implementation problems; (2) relatively broad and extensive consultation with both affected interest groups and implementing agencies; and (3) relative stability in the role specified for regional and local government in the implementation of national policy.

Structurally, policymaking and implementation are more highly differentiated in Sweden than in the U.S. The traditional view of Swedish policymaking and administration is that this differentiated system operates to produce a high level of consensus. As we shall see later, though, the level of consensus may be eroding. In addition to the distinction between legislative and executive functions that holds in the U.S., the executive function in Sweden is further differentiated into ministries {departement), with major financial and policy responsibilities, and administrative boards (centrala ämbetsverk), which carry out the actual administration policy from the national level (Anton 1980, 9-10, Hancock 1972, 205-206). This differentiation has two main effects: It locates policymaking functions in a highly visible set of institutions that are directly linked to parliament (Riksdag), and it creates a semi-autonomous base of institutional authority for administrative agencies. One might expect this greater differentiation to result in more conflict over policy and less attention to implementation questions in policymaking. In fact, the result has historically been the opposite.

The reason for this seeming paradox is that there are strong cultural norms and constitutional processes in Sweden that require extensive formal consultation among all parties affected by a policy before it is enacted, effectively stabilizing and containing potential conflicts. The two main structures of consultation are Royal Commissions, expert advisory bodies on key policy questions, and the remiss process, an established routine by which the ministries solicit detailed comments from administrative boards and interest groups on policy as well as

on implementation questions prior to making recommendations to parliament (Hancock 1972, 158). Remiss allows the government 'to get a feeling for the intensityof stimulated by a proposal... as it tries to determine what concessions,if should be made to protesting groups'. Proposals submitted for legislative decision after remiss 'include generous portions of the research on which proposals were based, a lengthy review of the remiss responses from affected organizations, and a consideration of how the proposed legislation has taken account of criticisms' (Anton 1980,7). In most circumstances, the result ofremiss is that 'by the time a legislative proposal is actually drafted, affected interests have been given a voice, potential opponents coopted if possible, and attention has been paid to implementation issues such as "who will carry out these proposals" "on whose budget" "do they want or can they be induced to do this" ' (Ibid., 7-8).

This structure of consultation is reinforced by the organizational structure of Swedish political parties and interest groups. Parties with parliamentary representation numerous, representing subtle gradations of ideology and economic interest, but virtually all have an organizational structure that closely mirrors thai: of national and local government, 'consisting of local and district associations, national committees, and small executive committees' (Hancock 1972,109-110). Furthermore, individuals attach more significance to party identification in Sweden, and their political activity is more likely to be expressed by participation in local party organization. The effect of this structure on legislators at the national level is that there are very weak incentives to engage in entrepreneurial behavior, since their electoral success is determined more by effectiveness of party organization than by the benefits they personally can deliver to constituents. The same organizational incentives apply to interest groups. 'Sweden is one of the most organized democracies in the world, both in terms of the number of special interests, that have created formal structures to promote their interests, and in terms of the extent to which these various organizations are centralized at the national level' (Anton 1980,12). The benefits that interest group leaders gain for their members are received through participation in structured consultation, rather than through individual entrepreneurship. The presence of an effective organizational extending from the local to the national level means, for both political parties and interest groups, that 'binding commitments can be arranged by very few people' (Ibid., 12).

Relations among policymakers and administrators are based on norms and institutions that provide a strong role for administrative expertise. The traditionalistparadigm' legislative-administrative relations in Sweden holds that Parliamentshould societal interests and that civil servants should have independent authority to administer programs', which can function to 'guarantee citizens against legislative excess'. Only recently has this paradigm begun to give way to the 'activist paradigm', in which 'administrators ... should do what they are told, ... Parliament should represent interests, and ... administrators and

politicians ... in the ... ministries should set policy' (Ibid., 100). Still, in spite of the activist paradigm, career administrators are viewed as a stable source of expertise with an independent base of authority, whose opinions are an essential ingredient in policymaking. Good administrators are viewed by their peers as "soft activists' - people with 'a sense of purpose, awareness of the opportunities available in the environment, the ability to sell both himself and his plans, aware of the needs of his subordinates, willing to participate rather than dictate, sensitive to the problems of his peers (and) above all, able to get along with people' (Ibid., 75). In a system built on norms of consultation, those with the 'soft' skills excel.

The role of local government in Sweden is based on both specific authority conferred by national policy, analogous but far from identical to national policy toward state and local government in the U.S., and also on a more general authority, which has a long constitutional and informal tradition in Scandinavia. The general authority specifies that municipalities shall 'conduct their own affairs', meaning that they are empowered to take any appropriate action that is in the interest of their inhabitants. Functions covered by this general authority include streets, parks, cultural activities, sports, leisure, water, electricity, heating, etc. Through specific legislation the national government calls upon municipalities to implement policies in areas such as social welfare, education, and land use (Gustafsson 1980). When Swedish municipalities implement policies based on such specific authority, they are not only empowered to add detailed prescriptions to national policy, they can also invoke their general powers to levy income taxes and charge fees to increase the quantity and quality of services prescribed from the national level. Currently, more than 45% of municipal expenditures are funded from locally generated income taxes. Since the organizational structure of political parties extends to the local level, local councils are constituted from party members who carry a strong identification with national parties. Hence, local implementation of national policy can be translated into a national party issue through the use of party structure. In addition, associations of local and regional governments, as well as associations representing the professional and non-professional employees of those governments, are among the most influential interest groups at the national level, which creates an additional link between national policy and local implementation.

The effect of these structural linkages from national to local government is a distinctive amalgamation of centralization and decentralization. The Swedish system is highly 'centralized', in the sense that political parties, interest groups, and - to a degree - administrative agencies have strong vertical organizations, and there are strong norms of consultation at the national level which promote a high level of consensus on national policy. But the system is also highly 'decentralized', in the sense of granting considerable autonomy to administrative agencies and municipalities shaping national policy to variations among localities.

This incentive structure has a number of distinctive effects on the content of national policy in Sweden. Broad consultation in the formulation of policy results in vague goals and diffuse proposals covering a wide range of issues. Furthermore, the political incentive structure encourages a relatively long time perspective, Swedish policy is typically prospective and preventive in its intent, rather than retrospective and remedial. That is, policies are designed more to anticipate and respond to developing trends than to repair existing problems. Prospective and preventive policy is difficult to make, however, since no one understands with any degree of certainty what will happen many years hence. The most rational response to uncertainty of this type is to leave the content of policy relatively unspecified, but to specify in some detail the process by which policy will be adapted to unanticipated events. National policy in Sweden, therefore, heavily on implements that specify patterns of consultation and process of planning at lower levels of government. The characteristic form of national policy is 'frame laws', or 'meta-policy', that mobilize the expertise of administrative boards, the general authority of municipalities, and the political influence of parties and interest groups behind institutionalized process of consultation planning (Gustafsson 1983).

This form of policy is well-suited to a governmental structure which is centralized the formation of policy but highly decentralized in its implementation. Ministries can focus on budgeting functions and specification of patterns of consultation, Parliament and the agencies engaged in prospective and preventive problem-solving. Municipalities can exercise a high degree of discretion, the bounds of their general authority, but their attention is focused on national policy by processes of planning and consultation. 'What really distinguishes Sweden from other systems', in the minds of its decisionmakers, is not technological advancement, the welfare state, or societal wealth, but 'a peculiar combination of social harmony within a dense organizational field - a combination sometimes referred to as "organization Sweden" ' (Anton 1980, 103), or as the poet Robert Frost once said of socialism, 'crowding with kindness'. This approach to problem-solving reflects a deep commitment to consensus and an equally deep aversion to open conflict. 'Swedish public institutions represent nothing so much as an elaborate structural labyrinth designed to prevent the; occurrence of open conflict, channel it if it does occur, and ritualize its consequences 160). The view of policymaking as prospective and preventive, and of implementation as structured consultation, is consistent with a broader 'pattern of seeking a clear definition for a limited, highly-specialized activity, creating a position to carry on that activity, and then permitting that activity to go on without further interference' (Ibid., 162).

National policies requiring collaboration across levels of government - in such
sectors as land use, housing, child care, care for the elderly, and employment -
take a distinctive form in Sweden. They typically define the policy problem,

describe specific authorities conferred on localities to deal with that problem, describe a structured process of planning at the regional and local level by which decisions are to be made, and often describe the division of fiscal responsibilities among levels of government. Beyond these parameters, the actual manifestation of policies in local government actions is a combination of the general authority of municipalities, or regions, and local or regional planning. Some degree of variabilityin from one locality to another is assumed to be a logical consequenceof and general local authority, not a manifestation of noncompliance.Once basic framework of policy is established at the national level and the process of planning commences at lower levels of government, changes in policy are manifested more often through administrative agreements than through changes in the implements of the original frame law. Hence, national policy contains considerable flexibility, or 'give', in the short term, and relative stability in the longer term. Implementation problems seldom require changes in the content of law for their solution, once the basic authority is set in place; more often than not, these problems are resolved through consultation among levels of government.

Similar patterns can be found in Swedish regulatory policy. In the field of occupational safety and health, for example, national law prescribes very little in the way of objects of regulation, performance standards, or enforcement procedures. general, regulatory decisions are subject only to the requirement that they be publicized after the fac \ along with the names of the responsible officials, those present when the decision was made, and the names of any dissenters to the decisions. In practice, specific occupational safety regulations are framed by small committees, composed of key officials from the administrative board, labour representatives, and business representatives. Meetings of these groups are neither secret nor public, they involve no outside witnesses, no reports, no transcripts. Appeals from the decisions of the administrative board go to the Cabinet, not to the judicial system (Kelman 1981, 13-15). National policy sets the broad frame for regulation; implementation proceeds by structured consultation.

Contrast this with the way occupational safety policy is implemented in the U.S.: The legislation itself is relatively detailed in specifying objects of regulation, administrative responsibilities, and procedures. When a threat to occupational safety or health comes to its attention, the responsible agency initiates a recommended rule; an advisory committee of experts is convened to evaluate the proposed rule; the committee is required to hold open meetings and to hear testimony from all interested parties; after extensive hearings, the agency publishes the rule for broader public comment and prepares extensive inflationary and environmental impact statements; the rule is then subjected to an open public hearing, in which the contending parties have unrestricted rights to cross-examine witnesses; after these hearings, the final regulation is published, accompanied by

a full statement of the reasons why it should be adopted; finally, the entire process is subject to judicial review, which can force a reconsideration of the rule from the beginning (Ibid., 10-13). National policy sets a relatively specific framework; implementation proceeds by public, adversarial processes.

As Table 1 suggests, then, differences in implementation between Sweden and the U.S. can be traced from the incentive structure operating on policymakers to the content of policy, both in terms of broad objectives and implements; hence, to relations among national and sub-national governments, and finally to basic characteristics of implementation processes. Two important qualifications must be made to this analysis, however. First differences are relative, not absolute. On the policymaking dimension, for example, the comparison shows only that political incentives lead to a relatively greater emphasis on implementation questions in Sweden than in the U.S., not that the one structure promotes consideration such questions and the other doesn't. Likewise, on the dimension of characteristic implements, the comparison shows only a relatively greater emphasis on structured consultation in Sweden and on formal delegation, distribution, and rights in the U.S., not that certain implements are present in one and absent in the other. Hence, the process of comparison becomes one of assessing the relative importance of certain incentives and implements. This assessment entails a certain amount of judgment, and is likely to result in debate on the fundamental differences in politics and governmental structure among countries. This is, we think, a useful direction for comparative studies of implementation to take. Second, our analysis has thus far treated differences between and the U.S. as if they were static, when in fact both systems are undergoing significant changes. These questions of fundamental relations between and governmental structure and of changes in political incentives are the subjects of the reminder of this paper.

The Interaction of Policy and Governmental Structure

Table 1 implicitly suggests that one reason implementation may proceed differently in Sweden and the U.S. is that policy may be 'about' fundamentally different things in the two countries. If policy is essentially prospective and preventive in Sweden, and retrospective and remedial in the U.S., then we may be talking about different conceptions of the role of government when we compare the content of policy in the two countries. Policies that appear to have similar subjects may be based on fundamentally different operating principles. Likewise, the characteristic implements of Swedish policy - structured consultation planning - are consistent with a governmental structure that assumes a basic compatibility of interests among levels of government, while the characteristic implements of U.S. policy - delegation, distribution, rights, and adversarial procedures - are consistent with a governmental structure that assumes a basic conflict of interests among levels of government.

Americans are legatees of the English tradition of social-contract liberalism, articulated by John Locke, in which government's justification derives from the necessity to protect the rights and property of individuals. The individual, 'defining and pursuing personal goals', precedes the state (Kelman 1981, 123). There is little allowance in this view for using government as an instrument to achieve the 'good society', since the good society is basically a result of the private pursuit of individual ends. Government exists essentially to keep people from misusing each other. Hence, any decisions to invoke governmental authority entails the Yebuttable presumption' that the collective good might better be served by allowing individual choices to run their course (Schultze 1977). In other words, the burden of proof resides with those who would use government to enforce collective ends. And, once a decision has been made to intervene, the burden of proof resides with those who would use public organizations accomplish collective ends, rather than private incentives.

Strong movements of progressive politics in late nineteenth and early twentieth century America have had to battle against and overcome the presumption that government intervention in economy and society were inimical to individual liberty. Liberal reforms, whether anti-trust laws, business regulation, social

security, or more recent attempts to combat poverty and provide income security, have all been justified in terms of expanding individual opportunity and liberty, rather than increasing collective welfare. Hence, the redistributive impulse of American domestic policy is weak. Policymakers assume that a healthy and expanding economy will provide for basic welfare, for equality of opportunity, and to a lesser extent, equality of condition. 'A rising tide lifts all boats', the sayinggoes.

The same basic principles apply to the relationship between national, state, and local government. If there is a rebuttable presumption against governmental intervention in general, then there is an equally strong presumption against intervention government the further removed it is from the individual. The structure of federalism promotes the idea, however inaccurate it may be from a constitutional that state and local government are somehow 'closer' to the people than the national government and that policies conceived at the state and local level have a presumptive claim to legitimacy for that reason alone (Beer 1978).

If one truly accepts the Lockean ideal of the protection of the individual as the primary, or even sole, goal of government, then it is logically impossible to give any content to an ideal of the 'public interest' that is more than the sum total of individual interests. The result is that American politics provides few, if any, overriding principles by which the competing claims of individuals and groups may be reconciled for higher, community values. Hence, when governmental intervention comes, it typically takes the form of specific solutions for highly visible problems - race, monopoly, environmental pollution, poverty - that have been forced on the agenda by temporary coalitions of reformers. The interests of groups, rather than of the collectivity, drive policy. Solutions are typically retrospective that is, they repair past inequities of externalities - and remedial - that is, they try to correct problems without fundamentally altering the fabric of society. Rights, adversarial procedures, and jurisdictional conflicts are built into governmental intervention as a way of limiting the reach of public officials into the lives of individuals and limiting the reach of one level of government into the affairs of another.

Sweden, on the other hand, is the legatee of a political tradition that fuses strong emphasis on individual competence with an equally strong sense of collectiveresponsibility family, community, and state. The state, in some fundamental sense, precedes the individual, rather than vice versa. This view stems, in part, from a pre-democratic society in which 'all sources of social power - political, economic, and religious - (were fused) into a single elite before which ordinary people stand both in awe and in fear and trembling', the so-called överhet state (Kelman 1981, 118-123). The process of democratizing the överhet state did not consist of asserting the rights of individuals against the prerogativesof but of 'domesticating' government to needs of individuals.Rather

viduals.Ratherthan viewing government as an enemy from whom key individual rights had to be wrested, it was viewed as a remote authority which had to be brought to the level of individual. The result is a conception of the state as a kind and helpful father, who is obliged to take care of social life as a whole. Apparent inconsistencies between individuality and collectivity that might seem obvious if the government was viewed as a threat to the rights of individuals disappear when a unitary state is domesticated to the level of the individual. The result is a "perception of the state as a service organization, rather than a mystical entity demanding the subordination of the individual to abstract goals of statescraft' (Hancock 1972,36).

Likewise, state, society, and economy are simply different manifestations of the collectivity, rather than separate estates with fundamentally conflicting interests. In such a conception of the collectivity, the 'good society' is more than simply a collection of individual preferences; it is a good to which many individuals subscribe. When governmental intervention appears necessary, policymakers need not rebut the presumption that collective ends might better be achieved through private decisions. The interests of the individual are, in some fundamental the interests of the collectivity, and vice versa. These interests are simply manifested in different ways in government, society, and economy. The function of policy is not simply to provide remedies for problems resulting from individual choices, but also to anticipate and prevent such problems before they occur. In this sense, policy is prospective andpreventive; the impulse to redistribute is strong. This view of policy results in a kind of seemless web of connections among the public and private spheres, as well as between the interests of the individual those of the state. Swedes are 'cooperative individualists' - strong in their sense of individual identity, but equally strong in their group loyalty and willingness to cooperate.

If the boundary between individual and state is permeable, so too are the boundaries between national, regional, and local government. The notion that authority should be divided among levels of government to protect individuals, or that one level of government is 'more responsive' than another, is less important than the more practical notion that certain governmental functions are better performed at one level than another. Hence, one finds strong traditions of local control and central administration residing side-by-side, without apparent contradiction. national government relies heavily on regional and local government for decentralized administration of specific national policies, but local government also preserves a reservoir of general powers that result in distinctive differences among localities. Likewise, strong identification of citizens with local government has not prevented extensive consolidation and reform of those governments (Gustafsson 1980). Jurisdictional prerogatives, and conflicts of interest among levels of government, are less important than making governmental compatible with collective ends.

Given these differing backgrounds, it is fair to say that policy really is about fundamentally different things in Sweden and the U.S. In Sweden, it is primarily about the assumption and assignment of collective responsibilities. In the U.S. it is primarily about the specification of individual rights and carefully limited governmental intervention in private choices.

These differences are important in assessing the interaction of policy with governmental structure in the two countries. On the surface, it would seem that both Sweden and the U.S. confront similar problems of centralization and decentralization the implementation of national policy. Both have undergone significant increases in the role of central government in economic and social regulation as well as in social welfare policy. In Sweden, the growth began much earlier and moved much faster than in the U.S. Both countries have undertaken extensive decentralization. In Sweden, decentralization came with extensive local government reform and consolidation; in the U.S., it came in the less dramatic form of block grants and revenue sharing.

One could make a strong case that the underlying dynamic of centralization and decentralization is the same in both countries, and that the remedies to implementation problems in both countries are, therefore, essentially the same. The basic model is: Political pressures mount for national intervention, policies multiply, the administrative structure for the implementation of policies increases in complexity, and the number of groups with an interest in national policy becomes greater. At some point, the growing responsibilities of national government beyond the level where they can be efficiently handled, and the symptoms of 'governmental overload' appear in the form of 'excessive cost, ineffectiveness, overregulation' (Beer 1977,6). At that point, political pressure mounts to decentralize and reduce the complexity of governmental functions. During periods of growth in national policy, there is a tendency to use 'hard' implements - regulations, standards, enforcement; during periods of decentralization, is a tendency to use 'soft' implements - greater discretion, simpler funding mechanisms, less central control. This fluctuation between centralization and decentralization results from interactions between policy and governmental structure. As the complexity of political demands on central government increases, the structure of policies becomes more complex, and the necessity for heavier reliance on lower levels of government increases. '[T]he extension national power calls forth wider participation' from affected constituencies, wider participation in turn calls forth greater political pressure to adapt national policy to regional and local variations (Ibid., 14). The interaction of policy with governmental structure, in other words, creates a need for 'softer' implements and 'softer' implementation strategies.

As a general model, this is a useful way to think about the relationship betweenpolicy
and governmental structure. But if our conclusion is
correct that policy is about fundamentally different things in Sweden and the

U.S., then the model does not go far enough. When policy means different things, centralization and decentralization also mean different things, and when centralization and decentralization mean different things, the movement from 'hard' to 'soft' implementation strategies has different consequences.

Because policy in Sweden is fundamentally about the assumption and assignment collective responsibilities, the interaction of policy and governmental structure is more likely to take the form of 'accommodationist' (Kelman 1981, 198) policies, implements, and strategies. Centralization, in this context, means assuming collective responsibility at the national level and allocating that responsibility or less rationally among levels of government. Decentralization means increasing reliance on lower levels of government to implement collective responsibilities in the main defined at the national level. In the U.S., where policy is essentially about the specification of individual rights and the limiting of governmental the interaction of policy and governmental structure is more likely to take the form of'adversarial' (Ibid., 198) policies, implements, and strategies. in this context, means intervention from the national level to remedy damages done to individuals by lower levels of government and private choices. Decentralization means increasing deference at the national level to decisions taken at lower levels, leaving questions of individual rights increasingly to the market and to state and local government.

Hence, the move to so-called 'softer' policies, implements, and strategies in both countries means somewhat different things. In Sweden, it means greater reliance on the relatively stable processes of collaboration and accommodation among levels of government, and greater reliance on informal negotiation and persuasion rather than explicit rules and procedures. In the U.S., it means relatively dramatic shifts in policy objectives and a withdrawal of the national government from the role of protecting individual rights in certain areas. In Sweden, 'softer' means greater reliance on processes of accommodation among levels of government; in the U.S. 'softer' means less involvement of the national government altogether in the decisions of private individuals and lower levels of government.

Furthermore, the two systems respond to the inevitable uncertainty and complexity that attends policy implementation in different ways. In Sweden, increasedlocal is the necessary by-product of a large public sector. Because the quantity of publicly provided services is so large and so many of them are distributed through local government, decentralization and local variabilityin provision of services is the only alternative to ungovernability. In addition, the structure of government never assumed, or institutionalized, an incompatibilityof between national and local government. Hence, the national government was relatively unconstrained in its use of'hard' implements as long as they worked. When the movement to 'softer' implements began, in response to increasing complexity and overload at the national level, the

transition was relatively smooth because it was based on long-standing norms of
accomodation and local variability.

In the U.S., on the other hand, the movement to increased state and local discretion is based only partly on the practical problems of administrative: complexity at the national level. The remainder of the explanation lies in the jurisdictional conflicts built into the American federal system. As noted earlier., increases in the size of national government were far less than proportional to increases national expenditures during the period of maximum growth. Also, the; U.S. public sector accounts for a relatively modest share of GNP, as compared with Sweden. Pressures for increased state and local control grew up as much in reaction to the encroachment of trie federal government on state and local autonomy, and to the perception that the federal government had gone too far in constraining individual choice, as in response to the necessities of administrative: complexity. Rather than a smooth transition from 'hard' to 'soft' implements, then, the response to increased complexity in the U.S. has taken the form of conflict and bargaining over which jurisdictions can and should assume which functions. The result is radical instability in the role prescribed for national government. Debates over the content of U.S. educational policy, for example., take the form of whether individuals should assume more responsibility to protect themselves from the uncertainties of the marketplace and whether the public: responsibility should be a state or a national function. In other words, responses to complexity and uncertainty in the U.S. have taken a form consistent with the: essentially adversarial nature of the governmental structure, in marked contrast to the accommodationist responses in Sweden.

Both Sweden and the United States, then, are grappling with the consequences; of increased demands on central government and both have responded by resorting 'softer' policies, implements, and strategies of implementation. In this; sense, the two countries are very similar in the way they have adapted to the complexity and uncertainty that accompanies implementation of national policy in the two countries. There the similarity ends, however. Accommodation to complexity and uncertainty occurs in Sweden against the background of strong accommodationist institutions growing out of a conception of policy rooted in a sense of collective purpose. In the U.S., this accommodation occurs against a background of adversarial institutions growing out of a policy working on the: margins of individual choice. In the first instance, the result is a heavier reliance: on informal negotiation and adjustment of implements consistent with the general purpose of policy. In the second instance, the result is instability in policy objectives, marked shifts in the implements used to carry out those objectives, and uncertainty over the roles of lower levels of government. In Sweden, reliance: on 'softer' implements tends to dampen uncertainty over the meaning of national policy; in the U.S., reliance on 'softer' implements tends to amplify uncertainty.

Changes in Political Incentives, Changes in Implementation

These differences, one must quickly add, are relative, not absolute. While the devolution of authority and control from national to sub-national governments seems to have occurred much more smoothly in Sweden than in the U.S., and with much less uncertainty about the objectives of the national government, both countries face significant changes in the political incentive structure that pose serious potential problems for the implementation of national policy.

In Sweden, the deliberative, open, rationalistic style of policymaking exemplified the remiss process has been substantially changed by the sheer number of separate interests which must be consulted, the number of discrete issues forced onto the national agenda, and increases in formal participation requirements. The national policy agenda has become so crowded and the number of interests affected so great, that some important issues are not subjected to remiss, while others are decided first among politicians and then subjected to remiss.

With increasing formal participation has come increasing 'sectorization', in which structures for consultation on the content of policy in such areas as health, education, transportation, and employment become increasingly specialized in membership and expertise. 'Rational coordination' among sectors at the parliamentary and ministerial levels has given way to 'bargained coordination' among sectoral interests. Lay advisory committees at the level of administrative boards are increasingly composed of union representatives, parliamentary members, and directors of closely allied boards, with the result that political questions are argued and re-argued well after the point at which they are nominally decided by parliament (Gustafsson & Richardson 1979, 423-429).

The traditional preoccupation with fact-finding remains, but it is no longer concentrated at the policymaking stage; it extends well into implementation. Complex policy issues are set into frame laws before they are fully aired at the national level, then supplemented with rules requiring extensive consultation and planning during implementation. The result is that consensus on fundamentals is postponed and important policy questions are exported, or pushed out, to the regional and local levels.

After more than 40 years of uninterrupted dominance by the Social Democratic Party, the government changed hands in 1976 and then returned to the Social Democrats in 1982. While this brief period of instability is nothing like the ebb and flow of American party politics in congressional and presidential elections since World War 11, it is a significant change from past political stability in Sweden.

Swedish ministries appear to be getting more involved in what were thought traditionally be administrative matters, and administrative boards appear to wield more influence in policymaking. Rather than signalling a clear redistribution of power, these changes seem to be more a blurring of the old differentiated system.

The overall effect of these developments is a weakening of incentives for policymakers at the national level to be concerned about whether policy is implementable, a higher degree of uncertainty both in policymaking and implementation, and considerably less agreement on the responsibilities of national, regional and local government. The political demands for consultation are so broad that it is more difficult to get agreement on specific policies; the risks of'immobilisme' are higher. Policies from the national level become increasingly 'symbolic', specifying issues and processes but not national objectives of the effects expected at lower levels. Still, the basic institutions and norms that promote attention to implementation questions are in place, and the patterns of consultation that give policymakers an incentive to pay attention to implementors are more or less intact.

The most significant change in the U.S. is the advent, in 1980, of a presidential administration that is less committed to devolving responsibility for implementing national policy to lower levels of government than it is to divesting the national government of many of its domestic responsibilities altogether. The budget and policy decisions of the early Reagan administration issued a strong warning to both the vertical and horizontal constituencies of past policies that the federal government would play a significantly reduced role. Block grant and revenue sharing proposals, although they figured prominently in the Reagan program, were not unique to that administration - they had been part of national policy in key domestic areas since the mid-1970'5. What was unique was the attempt to use expenditure controls and changes in legislation to dramatically reduce the federal role in many functions and increase private sector participation in others;. Significant expenditure cuts were made in national programs, which in some cases had already been changed to a block grant mechanism in the 1970'5. The diffuse and loosely-organized issue networks, which had formed around earlier federal policies, suddenly found themselves either defending the very existence of policies they had previously taken for granted or focusing their entire attention on budget questions rather than policy questions. The Reagan administration deliberately chose to make the existing gulf between political appointees and career officials even wider than it previously had been; its main objective was to reduce the size of the federal government, not to make it work more efficiently.

The effects of these changes have been threefold: (1) they have increased the perception on the part of state and local officials of instability in national policy; (2) they have dramatically reduced the already low level of consultation between policymakers and implementors; and (3) they focused the attention of people at all levels of government almost exclusively on changes in budget and policy rather than on problems of implementation. They have, in other words, reduced a previously weak incentive structure for attention to implementation to an even weaker one (Peterson 1984, Chubb 1985).

Both Sweden and the U.S. seem to be gravitating toward 'symbolic' policies -

Sweden as a solution to an increasingly cumbersome and unwieldly system of consultation and accommodation, the U.S. as a means of removing the national government from spheres where it previously had been. But the weakening of incentives to attend to implementation problems is more radical in the U.S. than in Sweden.

In both Sweden and the U.S. control of public expenditures has come to dominate, at least temporarily, questions of substantive policy and administration. In Sweden, however, the basic structures of accommodation and consultation that characterized the earlier periods of concern for policymaking and implementation still intact. In the U.S., the more diffuse structure of issue networks that supported what little concern was manifested for policymaking and implementation in disarray. By comparison, prospects for sustained attention to implementation questions are much better in Sweden than in the U.S.

REFERENCES

Allison, G. 1980. "Public and Private Management: Are They Fundamentally Alike in All Unimportant
Setting Public Management Research Agendas. Washington, D. C: U.S.
Office of Personal Management.

Anton, T. J. 1980. Administered Politics: Elite Political Culture in Sweden. Boston: Martinus
Nijhoff.

Beer, S. 1977. 'Political Overload and Federalism', Polity 10, 5-17.

Beer, S. 1978. "Federalism, Nationalism, and Democracy in America', American Political Science
Review 72, 9-21.

Chubb J. E. 1985. "Federalism and the Bias for Centralization', in Chubb, J. E. & Peterson, P., eds.,
The New Direction in American Politics. Washington, D.C.: The Brookings Institution.

Elmore, R. & McLaughlin, M. 1982. 'Strategic Choice in Federal Education Policy: The Compliance-Assistance in Lieberman, A. & McLaughlin, M., eds., Policy Making in Eudation. of the National Society for the Study of Education. Chicago: University of Chicago Press.

Elmore, R. 1982. "Social Policymaking as Strategic Intervention', in Seidman, E., ed., Handbook
on Community Intervention. Beverly Hills, CA: Sage.

Fenno, R. 1973. Congressmen in Committees. Boston: Little Brown.

Fenno, R. 1978. Home Style: House Members in Their Districts. Boston: Little Brown.

Gustafsson, G. & Richardson, J. 1979. 'Concepts of Rationality and the Policy Process, European
Journal of Political Research 7, 415-436.

Hamilton, A., Madison, J. & Jay, J. 1788/1948. 'The Federalist or, the New Constitution', in
Beloff, M. ed.. The Federalist. Oxford: Blackwell & Mott.

Hancock M. D. 1972. Sweden - The Politics of Post-Industrial Change. London: Holt, Rinehart &
Winston.

Hargrove, E. et al. 1981. Regulations and Schools: The Implementation of Equal Education for
Handicapped Children. Nashville, TN: Institution for Public Policy Studies, Vanderbilt University.

Hargrove, E. 1983. 'The Search for Implementation Theory', in Zeckhauser, R. & Leebaert, D.,
eds., The Role of Government in the 1980's. Durham, NC: Duke University Press.

Heclo, H. 1977. A Government of Strangers: Executive Politics in Washington. Washington,
D.C.: Brookings.

Heclo, H. 1978. 'Issue Networks and the Executive Establishment', in King, A., ed., The New
American Political System. Washington, D.C.: American Enterprise Institute, 87-125.

Kelman, S. 1980. 'Occupational Safety and Health Administration', in Wilson, J. O. ed., The
Politics of Regulation. New York: Basic Books.

Kelman, S. 1981. Regulating America, Regulating Sweden: A Comparative Study ofOccupational
Safety and Health Policy. Cambridge, MA: MIT Press.

Lynn, L. & Whitman, D. 1981. The President as Policy Maker. Philadelphia, PA: Temple University

Mayhew, D. 1974. Congress: The Electoral Connection. New Haven, CN: Yale University Press.

Murphy, J. 1973. 'The Educational Bureaucracies Implement Novel Policy: The Politics of ESEA,
Title I', in Sindler, A., ed., Policy and Politics in America. Boston: Little Brown.

Ogul, M. 1981. 'Congressional Oversight: Structures and Incentives', in Dodd, L. & Oppenheimer,
8., eds., Congress Reconsidered. Washington, D.C.: Congressional Quarterly Press.

Peterson, G. 1984. 'Federalism and the States: An Experiment in Decentralization', in Palmer, J. L.
& Sawhill, I. V., eds., The Reagan Record, An Assessment of America's Changing Domestic
Priorities. Cambridge, Mass.: Ballinger Publishing Co.

Rabkin, J. 1980. 'Office for Civil Rights', in Wilson, J.Q., ed., The Politics of Regulation. New
York: Basic Books.

Rosenthai, A. 1981. 'Legislative Behavior and Legislative Oversight,' Legislative Studies Quarterly
6, 115-131.

Salamon, L. 1981. 'Rethinking Public Policy: Third Party Government and the Changing Forms of
Government Action'. Public Policy, Vol. 29, 255-275.

Schultze, C. 1977. The Public Use of Private Interest. Washington, D.C.: Brookings.

Williams, W. 1980. Government by Agency: Lessons from the Social Program Grants-in Aid
Experience. New York: Academic Press.