Table Of Contents

• Introduction
• Allocation and Legitimacy from a Principal-Agent Perspective
• Mechanisms for Securing Allocational Legitimacy
• Allocational Legitimacy in the Global Economy
• Notes
  

Introduction

To what extent, in the name of promoting free trade, should the WTO encroach on the regulatory powers of its member states? This is by no means a new question for the international trade regime, but it has recently gained greater urgency. The Singapore meeting in November, 1996 will address this question in two ways. First, that meeting will explore how to further the objective put forth in the Uruguay Round of addressing the role of non-discriminatory national laws and regulations as barriers to trade. One issue is whether in the standards code for products or in the GATS for services, the post-Uruguay Round should go further than simply provide broad guidelines for policed national treatment, through the promotion of harmonization or mutual recognition of standards. Second, the Singapore meeting will take on so-called new issues, including the extent to which trade rules should be developed allowing states to impose trade measures on incoming products in the name of environment, labor, or competition standards prevailing in the home country. A corollary to this new agenda is the extent to which the WTO should endorse international standards developed in these areas through other international institutions or treaties. In both sets of issues, the extent to which WTO should centralise policy authority or leave it decentralised in the member states needs to be addressed with an eye toward the growing concerns worldwide for the erosion of sovereignty in an increasingly globalised economy.

The question of the WTO's authority fits into a more general debate over centralisation and decentralisation. Concern about the proper allocation of governmental authority has a long history in federal states, such as in the United States, and has emerged as well in European discussion of unification. Such concern centers on "whether centralization is preferable to decentralization, consolidation better than fragmentation, larger is better than smaller."⁽¹⁾ At the core of economic theories of federalism, particularly applied in the context of the United States, traditional analysis of the political economy of assignment have sought to determine the "optimal" distribution of public economic functions between different levels of government and the appropriateness of their possible transfer.⁽²⁾ Debate over the federal-state division of policy authority in the U.S. has returned with increased vigor in the wake of the Republican's 1994 sweep of Congress and several recent decisions of the U.S. Supreme Court. Similar questions also arise in the European context over the subsidiarity provision of the Maastricht Treaty, which expressly calls for a consideration of the level at which policy decisions can be "better achieved."⁽³⁾ While the precise terms of debate in the U.S., European, and WTO contexts undoubtedly vary, the core issue remains the same: What is the optimal way to allocate authority between upper and lower level institutions?⁽⁴⁾

In principle, both centralisation and decentralisation have their advantages.⁽⁵⁾ Decentralisation can allow government to respond more effectively to variations in local needs and preferences; to lower costs of planning and administration; provide opportunities and incentives for policy innovation; and give citizens greater choice and voice in policymaking. Centralisation, on the other hand, can permit government to address problems having cross-border (or spillover) effects; protect consumers against product risks; exploit available economies of scale; coordinate policies more effectively; and promote equality and political homogeneity across a larger domain to reflect "shared values." Centralisation may also secure a level playing field for fair trade, as states seek to limit "regulatory competition," although this argument is more contestable. It should be apparent, however, that the theoretical advantages of centralisation and decentralisation counterbalance each other, suggesting that the precise nature of an optimal allocation of authority may prove to be highly context-dependent. It may also suggest that optimality would be better conceived in a dynamic sense, allowing for changes in policy allocations over time.

Allocations, after all, do change. In order to exploit the advantages of centralisation, decentralised units of a tiered regime⁽⁶⁾ may delegate authority to a centralised unit. Similarly, the centralised unit in such a regime may at some point devolve authority back to the member states when decentralisation seems advantageous. In either case, the governmental level that currently possesses authority in a particular policy domain may seek to shift that authority to a different level. An allocational shift may arise for several reasons. The original allocation could have been mistaken, with the advantages anticipated from that allocation never materializing or other unintended consequences arising. Or circumstances may have changed in such a way as to necessitate a change in the allocation. For example, as long as air pollution remains a limited and localized problem, decentralised policy units can adequately address it. As pollution increases and spills over into other states, the issue becomes more appropriately addressed by centralised institutions.⁽⁷⁾ Changes in technologies, governmental capacities, information, and ideology may all prompt changes in authoritative allocations.

By what mechanisms are allocational shifts carried out? How can these shifts be secured in such a way to ensure that the transfer of authority from one unit to another will be used in the way it was intended? How do lower levels of governance retain some degree of control over higher levels in making centralizing allocational shifts, and vice versa? These are the central questions motivating our paper. We approach the centralisation-decentralisation debate through the lens of principal-agent theory, identifying the mechanisms by which governmental units can help ensure that the authority they delegate to another level gets used for the purposes underlying the delegation (and not usurped or abused for other purposes). After exploring the implications of principal-agent theory for issues of intergovernmental structures, we examine two different tiered regimes -- the United States and the European Union -- for evidence of the mechanisms suggested by our theoretical analysis. We argue that federal systems of any kind feature several characteristic mechanisms for sustaining the legitimacy of allocations of authority between different levels, and that policy debate surrounding proposed allocational shifts needs to include attention to these mechanisms along with the discussion of the appropriateness of centralisation or decentralisation.

Our comparative examination of the US and EU is relevant to the debate over the WTO. The problem in the WTO context is analytically much the same as in the US and European federal systems: "The title 'federalism' can be applied to any political structure in which the power of political authorities extends to less than the size of the economy in which resource movement is free of trade barriers. Its competitive action works within national economies in decentralised federal states or among countries associated in common-market arrangements with only limited central power."⁽⁸⁾ While there are obvious limits to drawing parallels between such divergent regimes, in each the institutions which issue rules for free trade and regulate goods and services often do not operate at the same level. Free trade places national governments in the same positions vis-à-vis WTO as individual states vis-à-vis a national US government or supranational EU structure. In each case, the federal institutions represent the collective action of member states and may take decisions which become binding on those states.

Attention to mechanisms for securing allocational shifts may well be more important in the WTO setting. Concerns about centralisation should be more pronounced in multilateral than in formal federal structures because the latter exhibit closer cultural, territorial, and economic ties that may lessen the need for controls by one level of government on another.⁽⁹⁾ Thus, to the extent that the U.S. and EU rely on mechanisms to address principal-agent problems, these mechanisms should be all the more relevant to the multilateral level.

Furthermore, as the autonomy of lower level units in federal or quasi-federal systems has become eroded, the US and EU have experienced a pendulum effect not yet observed at the global level. If the first direction of the pendulum is that of the progressive transfer of authority away from states in the name of free trade, the second direction is the progressive deceleration or even reversal or this transfer in the name of diversity and autonomy of constituent units. In experiencing the second direction of the pendulum --under the label of subsidiarity or devolution -- federations have developed rules and procedures that may be relevant to those systems moving in the first direction. Transferring the experience of federations to the WTO level may help to embed within the initial centralizing phase, tools that will make a subsequent, decentralising phase less necessary or less disruptive. In this way, our aim here is to say something about how subsidiarity might be secured at the global level.⁽¹⁰⁾

 

I. Allocation and Legitimacy from a Principal-Agent Perspective

A conceptual model to frame our understanding of the relationship between member states and upper-level institutions in tiered regimes can be found in the "new economics of organizations" exploring the characteristics of arrangements between economic actors that cannot rely entirely on market forms of control. Scholars have long noted that when one individual relies on the action of another, an agency relationship arises with common characteristics across issue areas. The individual taking the action is called the agent while the one affected by these actions is called the principal.⁽¹¹⁾ For this reason, the principal is supposed to define, monitor and reward the actions of his agent. The challenge in the principal-agent relationship arises from two sources: (1) differences in interests between agents and principals which lead them to prefer different goals and strategies; and (2) information asymmetries which come from the fact that "agents typically know more about their task than their principals do, though principals may know more about what they want accomplished."⁽¹²⁾ As a result, the agent may be able to perform tasks in ways which do not conform to the goals of the principal while the latter might not be able to do anything about it.

What, then, are the prescriptive implications of the agency approach? While agency tends to be viewed as a problem by economists -- for whom the issue is how to minimize unavoidable agency costs -- it tends to be viewed as a solution by analysts of organizations, or "a neat kind of social plumbing" to address the need for specialization, decentralization, and delegation.⁽¹³⁾ From both viewpoints, the principal-agent lens helps identify positive mechanisms to regulate the relationship between the two types of actors. In order to minimize agency loss, principals typically resort to a mix of inducement and enforcement methods, creating both systems of control or monitoring and systems of incentives. Systems of control allow the principal to devise ways to obtain information on the agent's actions in addition to the observed outcome, as well as to shape the agent's actions through the types of mandates the agent is given. Incentives are meant to do away with the need to gather full information and constrain the agent by aligning the agent's interests better with those of its principal. These mechanisms can be described most generically as different types of "agency ties," which form the ongoing web of connection between principals and their agents.⁽¹⁴⁾ More broadly and ultimately, as one author argues, "agency is more than a tie: it is a context for ties that cast shadow over commitment."⁽¹⁵⁾ Principals and agents are in a symbiotic relationship where the action of the agent commits the principal in ways that are supposed to have been shaped by that principal in the first place.

One of the core problems that the principal must face in engineering these relationships is that of the reversal of control where he comes under the control of the agent after the latter becomes what is termed a specialized purveyor.⁽¹⁶⁾ Among the challenges associated with the design of agency relationships is to avoid such reversal. The theoretical potential for reversal means that for some purposes it may be analytically more appropriate to treat an actor who originally or formally was designated the agent as in fact the principal.

How can the principal-agent conceptual framework be applied to the relationship between different levels of governance? And what are the specific insights that can be drawn from this framework regarding the mechanisms governing allocational control in tiered contexts? If we are to apply the principal/agent framework to explore the design of relationships between different jurisdictional levels, we first need to determine who is the principal and who is the agent in this context.⁽¹⁷⁾Any student of federalism will acknowledge that the issue is a complex and contested one. If we start from the premise that the principal is the actor who holds political legitimacy and has been entrusted with the "original" authority to act in a given sphere, we need to refer to the relevant legal delineation of powers or attribution of competence. Although we can assume that with any federal system, the "original" jurisdictional endowment lies with the lower level (the states) , constitutions or quasi-constitutions (if we consider the Treaty of Rome a onstitution-in-the-making) serve as mechanisms to allocate authority in alternative fields. We call the ultimate power that a principal reserves within a constitutional arrangement the jurisdictional endowment. In contrast, the power delegated by the principal to the agent we term the delegated authority.

Constitutions can be viewed as contracts that shift the endowment itself from the lower to the higher level of jurisdiction. Constitutions can also do the reverse. In the U.S., for example, many state constitutions explicitly shift authority originally endowed in the states to city or county governments acting as agents of the state. That the functions of principal can be split between levels is a core characteristic of federal system, as reflected in Wheare's widely accepted definition:


[In a federal system] powers are divided between a general level of government which in certain matters...is independent of the governments of the associated states, and, on the other hand, state governments which in certain matters are, in their turn, independent of the general government. This involves, as a necessary consequence, that general and regional governments both operate directly on the people; each citizen is subject to two governments.⁽¹⁸⁾



As a result, in cases where the upper level government acts in a given area, we need to specify whether, for the purpose of analysis, (1) authority has become reversed and the upper level is now acting as a principal (even though powers were originally delegated to it as an agent); or (2) the upper level is still acting on the basis of authority delegated to it by lower level units which continue to hold jurisdiction and define the upper level's mandate. Similarly, when we see states acting in an area, we need to ask whether they act because they have exclusive competence as principals or because they have been mandated to act by the higher level.

Up to now, we have spoken of agency ties in general to describe the relationships between the principals and their agents. More specifically, a number of insights can be drawn from the principal-agent framework to analyze the actual mechanisms governing allocational control in the context of tiered governance regimes. First, agency studies highlight the importance of strategic control, i.e., control connected with the formulation of the agent's task and overall objectives. The specificity of this formulation constitutes the initial constraint on the agent's margin of maneuver. A sales agent can be asked to act "in the best interest" of its firm, or it can be told to maximize the number of widgets sold in a given day. Similarly, as we will discuss below, the scope of an agent's delegated authority in inter-governmental contexts can be more or less specifically delineated by prevailing norms and specific legal acts.

Second, agency theories examine mechanisms of operational control, i.e., control connected to the on-going implementation of the task delegated to the agent. These consist above all in monitoring techniques that range from obligations of reporting regularly to the principals, to the standardization of task description and assessment according to specified categories. There may be a trade-off between such monitoring mechanisms and the first strategic aspect of control, in that the principal can afford to be flexible at the broad strategic level, if he can devise reliable modes of monitoring his agent's action. Similarly, in tiered systems of governance, principals may more readily delegate tasks to agents if they can subsequently monitor the implementation of the commonly agreed objectives by the latter.

There is a third, more radical, way for the principal to control the actions of the agent -- namely, by reducing its autonomy to the point where the principal shares in the activities of the agent. Thus, a joint venture with former suppliers strengthens agency ties by allowing the principal to become a party to the agent's operations. Conversely, we can also ask whether the agent has a say in shaping the mandate given to it. In other words, does the agent sit in the board room? In all cases, the question is the degree to which the principal and the agent actually share their respective functions, or participate in each other's deliberation and actions. This issue of power sharing is obviously crucial in governance contexts, where the degree of states' participation in federal structures reflects their level of influence or subordination. This is true whether we are concerned with the level of states' control in their capacity either as principals or as agents.

A final insight we extract from the agency framework is the question of the mechanisms by which the agency ties can be suspended or severed at any future time. Students of agency focus on the lengths of time of the agency relationship and the degree of entanglement implied. They also ask about the extent to which the actions or pronouncements of the agent commit the principal (does he act 'in the name' or 'in the person' of the principal?), and therefore the extent to which the principal can dissociate himself if he chooses to from his agent. As we will argue, the possibility of reversibility is the ultimate if blunt measure of control in inter-state arrangements.

Taking account of these various mechanisms, agency relationships in the business world can be arrayed along a spectrum from employee relationships (the strongest ties) to contracting out (the least strongest). Similarly, we will now analyze how relationships between different levels of governance can be arrayed along a spectrum defining the strength of agency ties between these levels according to the attributes of principal-agent relationships as outlined above.

II. Mechanisms for Securing Allocational Legitimacy

In making allocational shifts, one level of government gives up authority to another level. Agency ties provide some assurance to the principal that the authority it transfers to its agent will not be used contrary to the purposes of the delegation. The strength of the agency ties therefore defines both the power of the principal and the agent. The structural mechanisms that make up agency ties constrain the agent at the same time they reserve some power to the principal. In the context of tiered governance regimes, this means that attention to these mechanisms is necessary in order to analyze fully the allocation of authority between centralised and decentralised institutions.

We now turn to considering the specific agency ties that characterize principal-agent relationships in tiered governance regimes. We show how the four most characteristic mechanisms find their way into the established US federal system and in the EU's emerging federation. We cannot obviously cover in a single paper all or even most aspects of policy making in tiered governance regimes.⁽¹⁹⁾ Rather, we concentrate on the specific aspects of the US and EU allocational debates of most relevance to the WTO and which may otherwise illuminate the conceptual categories that we introduce.

A. Delineation

Delineation mechanisms provide for the determination of standards or guidelines that mark the scope of activity, functions, task or jurisdiction of the agent. These standards can be either jurisdictional or substantive; they can be more or less specific and more or less binding; they can circumscribe the domain of the agent through criteria for balancing different principles (e.g. trade vs. regulatory objectives) or through functional scope. While the respective powers of different levels are usually delineated in Constitutions or Treaties, residual powers -not explicitely enumerated- may rest with the constituent units (as in the US, amendment 9), the federal level (as in Canada, section 91), or the constituent units but with open-ended criteria for centralization. (as in the EU, article 235 of the Treaty of Rome).

One of the core powers traditionally delegated from the lower level to upper level has been that of regulating inter-state trade by policing the ways in which individual states at the lower level apply their own regulations to incoming goods, services and firms.⁽²⁰⁾ If the power to regulate products and firms remains at the lower level, the question that arises in this realm is what are the guidelines for delineating. This issue has engendered much debate in the context of the application of the dormant commerce clause by the U.S. Supreme Court⁽²¹⁾ and in that of the completion of a single market in Europe. The guiding principle for delineation has frequently been that of proportionality, or roughly balancing the burdens of intervention against the benefits.

The lower levels can also decide that laws and regulations themselves need to be centralized -- that is, designed and enforced at the upper level -- for reasons already discussed. In these cases, the scope of power of the upper level can still be limited by the extent to which it is allowed to constrain the lower level. One of the most radical examples of such limitations in the EU context has been to organize the delegation of authority horizontally at the lower level, that is from one member state to another, through so-called home country control, home country certification, or home country licensing based on the principle of mutual recognition. Here, careful delineation of jurisdictional authority between home and home state has served as an alternative to vertical delegation and centralization although sometimes requiring a minimum of centralized setting of minimum standards.⁽²²⁾

Another way in which the EU has sought to use delineation to apply the principle of subsidiarity has been through the increased resort to framework directives which open the way to alternatives such as voluntary agreements.⁽²³⁾ More generally, in the hierarchy of norms that characterize the EU legal systems, the least constraining norms (white papers, recommendations) are increasingly preferred over the most constraining ones (regulations). The Commission signalled this preference in 1995 with its SLIM (Simpler Legislation for the Internal Market) initiative to help simplify community directives, especially in the areas of construction and recognition of diplomas.

As already noted, in some instances the upper level of government in a federal system may be viewed as the endowed level (either through a reversal or acting in a secondary way, exercising its delegated powers legitimately).⁽²⁴⁾ When shifting authority back to the states, the upper level makes controlled or delineated delegations as well. For example, the federal welfare law in the U.S. allows states to depart from national standards in order to experiment with new approaches (the experiments are called demonstration projects). In order to undertake a demonstration project, however, the states must submit a detailed plan to the U.S. Department. of Health and Human Services (HHS). HHS reviews the plan under a set of substantive guidelines designed to ensure that the demonstration project shares the national aims of helping the poor and has appropriate evaluation methods in place in order to assess the success of the program. Guidelines like these -- and they are found in U.S. environmental law as well⁽²⁵⁾ -- serve to delineate the power being shifted from one level to another.

The application of delineating standards by the delegated level can be subject to review either by a different branch of the delegated level or by some branch of the endowed level, based on the pre-determined standards. So, for example, in the U.S. the general constitutional structure begins with states as the endowed level delegating power to the national level through explicit grants of authority in Article 1 of the U.S. constitution (e.g., interstate commerce, "tax and spend," and so forth.). These jurisdictional limits are enforced through review by the federal Supreme Court.

One critical issue is that of the balance between the judicial and political branches in assessing and mandating delineation. In the case of the EU, where the uncertainty about the extent to which the Treaty of Rome actually constrains states' domestic policies was great, the role of the ECJ shifted over time, in part as a function of the overall political context. The ECJ has been engaged in determining both the latitude of the lower levels to determine what criteria can justify their independent action and the extent to which itself as the upper level court can scrutinize these justifications. Its jurisprudence has progressively shifted in this regard, from the bold Cassis de Dijon jurisprudence in the early 1980s to the more cautious Keck and Mithouard jurisprudence after 1993. To some extent, the Court has taken its cue from debates at the intergovernmental level in order to ensure adequate delineation. Whether the ECJ should leave further fine tuning of the principle of subsidiarity to the political realm itself is a question open to debate, one which raises issues of power sharing discussed in some detail below.

B. Monitoring

While one level may decide to delegate authority to another, it may also retain some control by partially monitoring the other level's activities. Since a principal is unable by definition to monitor its agent completely, it may use proxies for such monitoring or require that the agent itself report in pre-determined ways.

Most monitoring occurs through democratic political process. Open debate, "government-in-sunshine" laws, and freedom of information laws all serve a larger purpose of making government accessible to the public (the ultimate principal in a democratic system), but these practices and laws also have the effect of facilitating monitoring of the upper level of government by the lower levels. State governments have organized collectively in the U.S. to create institutions designed, among other things, to monitor and influence the activities of the federal government as they affect the states. The Council of State Governments, National Conference of State Legislatures, and National Governors' Association have been instrumental, for example, in documenting the extent to which state governments have been burdened with complying with federal laws and regulations.

These state organizations were recently instrumental in securing congressional passage of the federal Unfunded Mandates Reform Act of 1995, an act designed to increase the amount of information about proposed federal legislation or agency regulations that would mandate compliance by state and local governments without accompanying appropriations to fund that compliance. Over the years, the federal government has adopted extensive environmental, civil rights, and labor regulations that apply both to state governments and the private sector. Although the private sector has borne most of the overall financial impact of such laws, it has become costly for state governments (especially in times of shrinking tax bases) to comply with federal standards. The new Unfunded Mandate Reform Act requires "that the Congressional Budget Office (CBO) estimate the cost of Federal mandates to State, local and tribal government" and that Congress take a special vote before adopting federal mandates which do not come attached to funding.⁽²⁶⁾ The Act serves to place some constraint on the federal government's use of delegated authority by infusing the legislative and regulatory process with information about the impacts of proposed mandates.

In the EU context, the greater decentralization of the implementation of EU directives has also been accompanied by a greater emphasis on inspection by the EU Commission. Monitoring is also relevant in instances where authority is delegated horizontally at the lower level, that is from one member state to another. So-called home country control, home country certification, or home country licensing based on the principle of mutual recognition, has been accompanied with rather stringent provisions for mutual monitoring in the EU context. By doing so, individual states are supposed to ensure that their delegation of regulatory authority is not open ended and is based on equivalent regulations enforced in the home country as they would have done in the host country.

C. Sharing

Sharing mechanisms involve representation or involvement by both levels of government in each other's affairs. They designate the extent to which the endowed units participate in the upper level decisions that will in turn affect them. This is different from allocations where the endowed level monitors, reviews, or approves the decisions of the delegated level according to some delineated standard. In certain areas there is no jurisdictional or substantive standard by which review should be made. Indeed, it is not delineation of power or review per se at all, but rather the design of structures at the delegated level that ensure that the endowed level retains some control over its actions. In addition to the issue of representation, sharing mechanisms include decision making rules (majority vs. unanimity) as well as the division of authority between the legislative and judicial branches in determining the standards for delineation of authority and the ways in which they should be interpreted and applied.

In the U.S. context, the federal legislative representatives draw from the states; presidential elections are based on the outcomes of state races; and states can prompt changes to the federal constitution. Under the original allocational contract -- the U.S. constitution -- the representational ties to the states were more direct than they are today. Prior to the 17th amendment in 1913, Senators were chosen by state legislatures instead of the voters of the states. As the U.S. Supreme Court has reasoned, "the principal means chosen by the Framers to ensure the role of the States in the federal system lies in the structure of the Federal Government itself. It is no novelty to observe that the composition of the Federal Government was designed in large part to protect the States from overreaching by Congress. The Framers thus gave the States a role in the selection both of the Executive and the Legislative Branches of the Federal Government."⁽²⁷⁾



The participation of individual states in upper level decision making is obviously still more pronounced in the EU context, although observers disagree on how extensive the involvement of member states should be. The EU, with its two headed executive (the Council and the Commission), and its two headed legislative branch (the Council and the Parliament), has one component of pure intergovernmentalism (power sharing among lower level components) and one of supranationalism (no lower level representation) in each of the two branches. The only branch which is purely supranational in the EU context is the judicial branch -- the European Court of Justice -- although even it is dependent on the national courts.

Even if they are represented at the upper level, states as principals may retain control collectively but not necessarily individually. Voting procedures at the delegated level therefore constitute a key control mechanism. The issue has always been at the heart of debates on state sovereignty in the EU context. For this reason, article 100a constituted the central and most controversial element of the Single Act negotiations in 1985. While it did not replace the original article 100 which had been the object of the infamous Luxemburg compromise on unanimity voting in 1965, it served to extend majority voting to all issues related to the internal market, "unless provided otherwise by the present Treaty," and with an explicit exception for fiscal harmonization, the movement of persons, and salaried workers.⁽²⁸⁾

D. Reversibility

When endowment remains at a given level within an allocational regime, any delegation of authority from the endowed level to the delegated level will be reversible to some extent. Any allocation will likely satisfy two apparently contradictory requirements: legitimacy and reversibility. Legitimacy demands both a sound justification for initial allocations of authority and a degree of permanence or "stickiness" to these allocations to ensure respect for allocational choices. Reversibility, in contrast, implies that initial allocational choices may need to change due to unintended consequences or subsequent changes in technologies, ideology, or circumstances.

In thinking about reversibility, two issues need to be kept in mind: (1) First, is there any reversibility left? If not, then the endowed level has given up all its authority and ceases to hold an endowment. (2) Second, how easy is it to reverse an allocational delegation? Here we distinguish between the reversibility inherent in any endowment (what we might call a reserved or implicit reversibility) and the more explicitly-stated reversibility (express reversibility). Even among express mechanisms of reversibility, however, it is best to think of these along a continuum, of either more or less, easier or harder reversibility.

Sunset provisions are among the clearest examples of reversible mechanisms. Here the delegation expires at a date certain and the authority automatically reverts (upon the tolling of the deadline) back to the endowed level. We find in the U.S., where the national government possesses a secondary endowment over issues such as welfare (tax & spend) or environment (commerce), frequent examples of sunset provisions. Welfare waivers granted by the federal government to the states to allow the states to experiment with welfare reform expressly state a time limit, usually of a few years or so, for the duration of the waivers. Similarly, the national approvals of state implementation plans for environmental laws have a limited duration, requiring states to resubmit new plans on a regular basis.

Revocability provisions are another example of reversibility mechanisms. These are "escape" clauses that permit the endowed level to step in to revoke the delegation under certain circumstances. Environmental laws in the U.S., for example, permit the federal government to revoke delegations made to states (with the federal government acting as principal) in public health emergencies. Many of the directives implementing the single market in Europe and involving some degree of mutual recognition include rather broad allowance for escape when states needed to "protect the public interest."

At another level altogether, national negotiators in Europe in the framework of the IGC 1996 are considering a clause that would allow states to exit the union under specific circumstances (even though it has increasingly been considered impossible to do so in practice). Similarly, if states in the U.S. remain the ultimate endowed level as they were at the founding of the constitution, we should therefore expect them still to have the (theoretical) ability to revoke any delegations of authority given in article 1 to the federal government. The amendment provisions in Article 5 of the constitution do give states the theoretical -- but heretofore unused -- option of calling for a constitutional convention on the vote of two-thirds of the state legislatures, which preserves an extreme option for states to control the federal government.

The EU has resorted to escape clauses, to buy off states' adherence to increased constraints put on their capacity to regulate domestically. A prime example is that of the Single European Act where the introduction of a safeguard clause constituted a clear quid pro quo for the extension of majority voting (article 100a.4). This provision allowed a member state to opt out -- e.g., to continue to apply its national provisions to imported products -- if it felt that its interest was overruled by a Council vote. This was a clear departure from the principle of "preemption of harmonization" established by the Court's jurisprudence, under which host country requirements could only be justified "in the absence of Community measures."⁽²⁹⁾ This clause led some prominent observers of the Community to regard the Single Act as a set-back in community integration.⁽³⁰⁾ To be sure, the scope of the safeguard clause was itself limited since it could not be used against a unanimity decision by the Council and needed to be justified "on ground of major needs," as provided for in Article 36 of the Rome Treaty, or relating to protection of the environment or the working environment. Ultimately, however, political pressures would be the strongest limit against the systematic use of the safeguard clause.

 

III. Allocational Legitimacy in the Global Economy

We have examined four mechanisms -- delineation, monitoring, sharing, and reversibility -- that often accompany transfers of policy authority in tiered regimes such at the U.S. and EU. At the outset of this paper, we stated that our aim was to move beyond the longstanding debate over the appropriateness of centralisation or decentralisation. Any inquiry into the proper allocation of policy authority, we have argued, needs to give attention to allocational mechanisms and how they are crafted, combined, and traded off against each other. Allocational mechanisms help prevent unintended consequences resulting from a shift in authority and in doing so help make those same shifts possible by providing assurance to the governance level making a delegation of power.

Within each type of mechanism, it is important to consider how much control the principal has left after authority has been transferred. As we have suggested, agency ties vary in their strength. Delineations can be narrow or broad; monitoring regularized or happenstance; sharing direct or indirect; and reversibility implicit or explicit. The variation in the strength of these mechanisms means that tradeoffs can be made between them. Narrowness in delineation, for example, might be sacrificed if sharing is high or reversibility easy and clear. Combinations of these mechanisms are also possible -- and indeed almost always do accompany allocational shifts. When the mechanisms are used in combination, the endowed level retains greater control over the actions of the delegated level.

The degree to which power is centralised (or decentralised) therefore depends on more than simply a (static) decision about the appropriate level for exercising policy authority. It also depends on how that authority gets transferred, and what mechanisms exist for modifying, constraining, and abandoning the transfer if necessary. If states make up the endowed level, and authority is being transferred to a central institution, then tightly constraining mechanisms preserve the (at least latent) advantages of decentralisation. The principle of subsidiarity, for example, can be used as a delineating norm if it is inscribed in a treaty or constitution, and if used in conjunction with monitoring, sharing, and reversibility can entrench a preference for decentralisation, even in a regime where authority gets transferred to a centralised institution.

The allocation mechanisms described here can be thought of as complementary and interconnected components of the overall degree of decentralisation in a given tiered regime. Together, they cover the different facets of subsidiarity as generally understood in the European context, including the limits and constraints put on the delegation of authority from the lower to the higher jurisdictional level;⁽³¹⁾ the requirements to decentralize the implementation and enforcement of tasks formulated at the higher level; and the formulation, monitoring and enforcing of the delegation of jurisdictional authority horizontally from one unit at the lower level to another unit.⁽³²⁾ They constitute a more generic framework of analysis than simply asking what powers are delegated to whom and can be used to think about allocational issues in more dynamic and process oriented ways. Several sets of questions follow from our analysis. First, under what conditions are we likely to see different kinds of allocational mechanisms in place? Are endowed national levels likely to act differently than endowed state levels? (We might think that sharing mechanisms are more likely with state endowed levels delegating to some higher level, where representatives from the states can serve on the decision-making body of the higher, delegated level. But do we ever see the reverse?)

Second, under what conditions are some mechanisms more appropriate than others? Does value of the strength of constraint depend on trust, shared cultural understandings, importance of the policy issue at stake, or the likelihood of error in making the allocation?

Third, in the context of centralising allocational shifts when do some states (say, larger or more powerful ones) need stronger allocational constraints than other states? When is it optimal to have differentiated degrees of constraints (such as the vetos reserved to permanent members of the UN security council)? Under what conditions are constraints better made equal across states or unequal? In the U.S. Senate, for example, there is a disproportionate influence of small states relative to their size. Maybe big states need smaller states more and will give them greater constraints. Of course, one could look at the disproportionate involvement of larger states in the House and their disproportionate impact on presidential elections and make the reverse claim: namely that larger states have the disproportionate power. In any event, the key question here is whether the constraints need to be -- or should be --equally shared when the endowed level consists of multiple, independent units -- and if not, then what is the optimal arrangement.

Finally, the concern arises that in allowing endowed states to rein in their delegations when serious mistakes become evident, these same states are also permitted the ability to change the delegation when the nature or seriousness of a mistake is not so clear. We may, after all, be mistaken about mistakes. Allocations might be changed for reasons having nothing to do with perceived mistakes, but rather with capture or rent-seeking behavior at the endowed level. Although we have treated lower level units most often as the principals in tiered governance regimes, they are themselves actually but agents of the public, and the risk may remain that these levels could shirk their responsibility to the public interest.

Many of the issues we have raised in this paper are becoming relevant to current debates in the WTO, where the set of "new issues" that will be discussed in Singapore in November 1996 give rise to the same needs for devising allocational control mechanisms as has been experienced in the US and EU. The current agenda is to attempt to reconcile the objective of the existing trade regime embodied in the WTO with both other international regulatory regimes in the field of labor or environmental standards and with current state-level policies and standards, including labor, environmental and competition law.

The current WTO agenda is guided by two rationales that bear on the issue of subsidiarity. The first, positive rationale is to lend some enforcement power to standard setting exercises at the international level and to ensure that existing trade-related clauses in non-WTO treaties be made compatible with WTO rules. The second, negative rationale is to ensure that the current legal vacuum at the international level does not leave authorities at lower levels of governance free to erect trade barriers in the name of unilateral enforcement of standards, a trend that could erode the international trade regime. Hence, states will be engaged in devising a new shift in allocational authority from the lower to the upper level of governance. At the same time, however, in a context where economic globalization is often perceived as a threat by normal citizens, the very legitimacy of these allocational shifts risks being questioned if states do not retain a significant degree of control over these policy area. We argue that while securing subsidiarity at the global level is an imperative that can be secured through similar mechanisms as in the EU and US contexts, the tradeoffs that can or should be made between these mechanisms may differ significantly at the global level.

The starting point facing the community of member states is to agree on mechanisms of delineation that will keep the upper level within the bounds ascribed to it and to ensure that the agreed balance is kept between state autonomy and collective goals, lower and upper levels of governance. This applies whether the upper level sets international minimum standards or whether it implements criteria for determining whether a domestic measure applied to a foreign products, firms or people should be considered legitimate. For a start, these delineation constraints apply to the way the WTO will implement the new agreement on Technical Barriers to Trade which supersedes the GATT Standards code. Among other things, the agreement not only calls for the use of the traditional least-restrictive-means principle, but also articulates broad principles for assessing whether a state regulation constitutes an unnecessary barrier to trade. Failure to use existing international standards when appropriate may constitute such an unnecessary barrier. How are these vague criteria going to be fine-tuned? Similarly while the GATS is supposed to ensure that qualification requirements, technical standards, and licensing requirements "do not constitute unnecessary barriers to trade," the agreement provides only general guidelines for doing so. Regulations applied to foreigners must respect the principle of proportionality and must be administered in a "reasonable, objective, and impartial manner." How transparent, accountable and legitimate is the process by which this is to be determined?

Looking ahead, if the WTO is to endorse new international standards in the fields of environment, labor or competition, how constraining should this endorsement be? Should it rely on flexible standards as a function of levels of economic or economic development? Should it indicate general performance guidelines or more specific ones? Devising appropriate delineation mechanisms to ensure subsidiarity will be key; but it may also prove easier to increase subsidiarity in tandem through the other allocational mechanisms analyzed in this paper. In this regard, it might be argued that reviewing the implementation of delineation standards at the global level should remain to a great extent under the purview of inter-state negotiations and political working groups rather than be left to judicial adjudication. Similarly, provisions as well as resources for mutual monitoring between states should be given greater attention. Finally, it may be argued that reversibility can constitute a much less controversial allocational mechanism at the global level than in the context of the US or the EU. Presumably formal withdrawal of membership from WTO and other standard setting organizations is much easier than in the case of the US and the EU. While this is not likely to be common in practice, broadly defined escape clauses may allow to states to justify temporary (or even permanent) exemptions from obligations and reassure participating states that they have not foregone forever their precious territorial sovereignty.

Notes

1. Susan A. MacManus, "Federalism and Intergovernmental Relations," in William Crotty, Political Science: Looking to the Future, vol. 4 (1991), p. 204.

2. These theories, which seek to provide a functional rationale for the division of federal vs state power to intervene in the economy, are usually constructed as decentralization issues for mature federations. As presented, for instance, by Oates, "the determination of the appropriate degree of decentralization for a particular government sector ... requires matching public functions with their appropriate levels of application." Wallace Oates, "An Economic Approach to Federalism," in Fiscal Federalism (Harcourt Javanovich, 1972), p3.

3. Treaty on the European Union, Article 3b, reads in full: "In areas which are not under its exclusive power, the Community shall act in conformity with the principle of subsidiarity, only if and in so far as the objectives of the proposed action cannot be sufficiently achieved by the member states and can therefore, by reason of the scale and efforts of the proposed action be better achieved by the Community." A similar test of where policy can be best attained is found in the The Single European Act's provision with respect to environmental policy: "The Community shall take action relating to the environment to the extent to which the objectives referred to in paragraph 1 can be attained better at the Community level than at the level of individual member states" (Article 130R).

4. The terms "devolution," "decentralization," "delegation," "federalism," and "subsidiarity" share a common concern with the proper allocation of authority in what we call tiered regimes. We use "tiered regimes" deliberately to include both nations organized along federal lines (e.g., the U.S., Germany) as well as inter- or supranational institutions comprised of nation-states as members (e.g., the European Union). Because we mean to encompass both kinds of tiered regimes, we use the terms "lower-level" and "upper-level" governments as the generic terms for units in either kind of regime.

5. For a summary of the advantages of decentralisation and centralisation, on which this paragraph draws, see Reiner Eichenberger, "The Benefits of Federalism and the Risk of Overcentralization," Kyklos 47:403-420 (1994).

6. Our use of the word "regime" to include federal and quasi-federal systems, as well as international institutions such as the WTO, may seem a bit unconventional. We do recognizing its conventional use, but also hoping to convey the similarities between the different kinds of governance structures. We do not mean to imply that important differences do not exist across these domains.

7. Sometimes actual changes in the conditions of the world do not drive allocational shifts, but rather changed knowledge about these circumstances do. The case of transboundary sulfur dioxide pollution is one such case. See Bill Clark, Project on Social Learning and Acid Rain (unpublished).

8. Jean-Luc Migué, Federalism and Free Trade, Hobart Paper 122, London: The Institute of Economic Affairs, 1993, p. 27.

9. For a discussion of the importance of shared cultural and territorial domains for federal structures, see Daniel J. Elazar, "International and Comparative Federalism," PS: Political Science & Politics 26:190 (1993).

10. In Europe, the principle of subsidiarity takes on at least two dimensions. The first involves the level at which authority should be vested for specific policy areas, and it mirrors the long-standing debate over centralisation and decentralisation. The second dimension arises once a decision is made to delegate authority to the center. This second dimension raises the question of how intrusive policy-making by the center should be, and is often referred to as the principle of proportionality or intensity of action. Commission of the European Communities, " The principle of subsidiarity" Communication of the Commission to the Council and the European Parliament", SEC (92), Brussels, 27 October 1992. Our approach encompasses these two dimensions, but also seeks to address a much broader question of how member states can retain some of their control over a policy area even after making a delegation to the center.

11. For the seminal overview, see John Pratt and Richard Zeckhauser, "Principals and Agents: An Overview" in John Pratt and Richard Zeckhauser, Principals and Agents: The Structure of Business, Boston : Harvard Business School Press, 1985.

12. Pratt and Zechauser, p 3.

13. White, p 188.

14. The spectrum of agency ties has been thoroughly explored in the business context, from the analysis of merchant networks to the running of large corporations. See Harrison C. White, "Agency as Control," in John Pratt and Richard Zeckhauser, Principals and Agents: The Structure of Business, Boston : Harvard Business School Press, 1985, p 187-209.

15. White, p. 189.

16. White, p. 205.

17. We acknowledge that in asking this question we are implictly treating governmental units as unitary actors. Although this assumption grossly misrepresents the workings of complex institutions and ignores the multi-level games that go on within them, we plead the need of modellers everywhere to make simplifying sacrifices for the larger cause (we hope) of analytical clarity.

18. K.C. Wheare, Federal Government, 4th edition, New York: Oxford University Press, 1964, p. 2.

19. The issue of optimal assignment obviously applies to the whole array of economic policy making functions, that according to Musgrave can be categorized under a triptych, encompassing, in his terms allocation, stabilization and distribution functions R. Musgrave & P. Musgrave, Public Finance, Theory and Practice , (New York: Mc Graw-Hill, 3d ed, 1980). Musgrave's functional classification is widely used in the EC context. See for instance the Padoa Schopa Report,Commission of the European Communities, Luxembourg, 1987. Note that we use the term "allocation" differently than Musgrave to refer to the assignment of policy authority to different levels of government.

20. See U.S. Constitution, article 1, section 8, clause 3 (the "commerce clause"); EU Treaty of Rome, articles 30 and 59; GATT articles.

21. See, e.g., Pike v. Bruce Church (1970); Hunt v. Washington State Apple (1977); Philadelphia v. New Jersey (1978); Hughes v. Oklahoma (1979); Raymond Motor Transportation (1980); New Energy Co. of Indiana v. Limbach (1988).

22. For a discussion see Kalypso Nicolaïdis, "Mutual Recognition of Regulatory Regimes: Some Lessons and Prospects," in Regulatory Reform and International Market Openness, OECD Publications, November 1996.

23. See, for instance, the review of the fifth environmental program.

24. Indeed, the only other work we know of which analyses U.S. federalism in principal-agent terms explicitly and only treats the federal government as the principal vis-à-vis the states. See John Chubb, "The Political Economy of Federalism," American Political Science Review 79:994 (1985). The U.S. Supreme Court has recognized that "the sovereignty of the States is limited by the Constitution." Garcia v. San Antonio Metropolitan Transit Authority, 469 U.S. 528 (1985). But cf. New York v. U.S., 112 S. Ct. 2408 (1992) andU.S. v. Lopez, 115 S.Ct. 1624 (1995).

25. See, for example, the Clean Air Act's provisions for state implementation plans. 42 U.S.C. 7407, 7410.

26. Senate Report No. 104-1, p. 5.

27. Garcia v. San Antonio Metropolitan Transit Authority, 469 U.S. 528 (1985)

28. The exclusion of the movement of people ( article 100a.2) constituted a political compromise with Britain who consistently argued that a common immigration policy ought to be a prerequisite to free movement of persons and refused to see immigration rules fall under a majority voting procedure --for fear of having to adopt Germany's more lenient rules of entry.

29. Under the Court's jurisprudence, after harmonization has been completed a member state cannot invoke Articles 36 for goods and 56 for services in order to avoid having to comply with a Council's decision.

30. See Pescatore, "Some Critical remarks on the 'Single European Act'," in Common Market Law Review 24: 9-18, 1987.

31. In the EU context, tand in areas of mixed competence, subsidiarity requires that action at the EU level must be based on an appaisal of how necessary the proposed measures are (can the objective be satisfactorily attained by member states) and how effective they will be (can the objective be better attained by action at the Community level?).

32. It is worth noting that, in the spirit of subsidiarity, mutual recognition had served to limit the transfer of sovereignty to supranational European authority. But mutual recognition is not actually a straighforward example of subsidiarity in that it consists itself in a delegation of authority between states.