Erik Moberg:

Competition in Government by Federalism: a Discussion of the Idea

ã Erik Moberg and the City University of Stockholm



1. Hierarchies and Federations
    Multilevel Decision-making
    Hierarchies in General
    Hierarchies for Governing
    Mixtures of Hierarchical and Federal Elements

2. The Unitary System
    The Border
    Legislation and Implementation by Hierarchy
    The Principle of Subsidiarity
    Risks for Bad Government

3. Federalism - Some General Aspects
    Mistaking Hierarchies for Federations
    Reasons for Two-Level Systems
    Problems with Two-Level Systems
    Fusion, Fission and Equilibrium

4. Competition in Law
    Competition in Policy
    Deficiencies in the Market Analogy
    The Conditions of Variety and Mobility
    The Framework for the Competition
    Competition in Law through Fission and Fusion
    The Creation of Good Societies




In this paper I will treat the two subjects of federalism and competition in government, and in particular I will focus on their bearing on each other. Certainly the two subjects may be treated separately, and that has also been done to a considerable extent. A lot may be said about federalism without mentioning competition, and a lot may be said about competition in policy without having federalism especially in mind. Still there are strong reasons for treating the two subjects together. Thus, governmental competition may be particularly beneficial when taking place within a federal framework, or at least that is sometimes argued. It should therefore be interesting to discuss that idea in some detail, and that is the main purpose of this paper. I will also try to make the discussion relevant to Swedish politics. That, as we will see, means addressing two main questions. The first one is about the idea of arranging policy competition within Sweden, by making the country federal in some way. The other one is about the idea of creating, for instance within the EU, some kind of policy competition in which many countries, among them Sweden, may participate.

1. Hierarchies and Federations

Multilevel Decision-making

It is a well-known phenomenon that public decisions, depending on their nature, are taken at different levels. Almost all nations do, for example, have a central, or national, level, as well as a level of local communes. Some nations also have a federal level, and a state level. Above the nations there is furthermore the possibility of a supranational level and there are, in the real world, at least some rudimentary signs of actual supranational decision-making. The existence of different levels is thus obvious, but in spite of this the basic principles behind these multilevel structures are hardly obvious and, in fact, not even well understood in the relevant social sciences. What is the raison d'etre of these structures? Which purposes do they serve?

A first important point is that there are different kinds of multilevel structures. A main distinction is the one between hierarchies and federations. Below, I will mainly deal with federations, but even so it is expedient to describe hierarchies carefully in order to make the distinction clear. Looked upon superficially the two types of structures are quite similar, and they are actually sometimes treated interchangeably, even in scientific texts. I will give some examples of that later on.

But even if the emphasis is on federations, I will start by describing hierarchies since they are easier to understand.

Hierarchies in General 1

Let us consider the task of building a big highway, within a scheduled time and according to specified criteria of quality, through a nation with varying climatic and terrain conditions. Clearly, this task requires the participation of a large amount of people, and obviously the activities of all these people must be coordinated in some way. For this type of operation a hierarchy, like the one in figure 1, is a fitting organization. At the summit there is a chief executive, who is responsible for the overall planning as well as for the division of the complete task into subtasks, such as the building of parts of the highway. Heading these sub-tasks are the functionaries at the level immediately below the chief executive. Similarly, these functionaries may have officers below them and finally, at the bottom line, are all the individual workers actually building the highway. Thus, the lower a level is, the more people it contains. Furthermore, at each level people are obeying orders from those at the level immediately above, but they are also given a considerable latitude for acting according to their own judgements. Such action, however, is expected to be in line with, and to serve, the common, overall goals, of the entire operation.

Figure 1

The example illustrates the basic purpose of a hierarchy. Assume, for instance, that all the individual workers were left on their own, without any superior command; then their behavior would be chaotic and uncoordinated, and thus ineffective. Assume, on the other hand, that the orders of the chief executive were utterly detailed and gave no latitude at all for individual action at lower levels; then the individual worker would be constrained, and without any possibilities to adapt his behavior to the needs of his individual situation, which only he himself is familiar with. So clearly, there is a lot to be achieved by allocating the decision-making to a number of levels, and by giving the people at each level a considerable freedom to act upon the particular information they happen to have at their own particular level. It is precisely this possibility that is provided for by a hierarchy. At the same time, however, it is important to note that the decisions taken at any level are always intended to serve the goals of the entire structure as formulated by the leadership at the summit. Basically, the hierarchy is a command structure. The reason for the delegation of decisions to lower levels is just expediency. The organization works better, with reference to its own goals, if the power to take decisions is distributed in a proper way to the functionaries at the different levels.2

The purpose of the hierarchy in the example above was to build a highway and there are, of course, countless similar hierarchies created for the execution of big and complex tasks. Such hierarchies are, for instance, omnipresent in industry. Even military hierarchies, which make big defensive or offensive operations possible, are of interest here. At the top of such an hierarchy there is a commander in chief with responsibility for the activities along the whole front line; immediately under him there is a small number of sub-chiefs, each one responsible for his rather large section of the front; and so forth down to the individual soldiers, each one responsible for just a few meters of the entire front. As in other hierarchies, considerable latitude is given to the decision-makers at each level. The individual soldier, for instance, decides himself about the exact pointing of his gun, and perhaps about the exact moment of shooting. What is interesting about the military hierarchy, in contrast to many other hierarchies, is that it has an enemy; and the enemy is likely to concentrate his efforts at those parts of the front where, whatever the reasons, the hierarchy is weakest. In a similar sense some governing hierarchies have enemies, for example citizens trying to escape a tax-collecting hierarchy.

So far I have taken for granted that the orders given in a hierarchy are transmitted effectively downwards to the functionaries they are intended for, and that those functionaries obey the orders completely. In reality, however, neither assumption is likely to be valid. First, orders will always, to some extent, be distorted or misunderstood and, second, the functionaries will always have some goals of their own and will therefore try to serve these goals rather then to implement the given orders completely. Together these two mechanisms put a practical limit on the size of a hierarchy. If it becomes big enough, for instance in terms of the number of levels, the lowest level will act independently of the orders given at the hierarchy's summit. This phenomenon is sometimes described as bureaucratic free enterprise (Tullock, 1965).

Hierarchies for Governing

Governing a country involves, among other things, the enforcement of laws. This task, in turn, contains the obligations of making the laws known to people, of supervising people's conduct, including obedience to tax laws, and of reacting when laws are broken. These tasks are reasonably more demanding the bigger the nation is (in terms of area, or population, or both). They are also more demanding the more reluctant people are to stick to the rules.

Let us assume, for a beginning, that we have a very small nation. If so, it seems quite likely that a single uni-level government for the whole country would serve perfectly well. Since the area and the population are assumed to be very small, such a government will be able to govern on its own. If, however, the nation is considerably bigger in terms of area and population, the governing tasks may be too demanding for a single government. In such a situation a hierarchy is a likely solution. The nation may be divided into a number of regions or districts, each one with a chief, who, as an agent of the central government-principal, is responsible for the law enforcement within the district. At some still bigger size it may be rational to divide the districts into sub-districts, each one with its own chief, and so on. Sheer size may thus justify a hierarchy for governing.

In reality these kinds of hierarchical patterns are omnipresent. It is common to divide nations into administrative districts, which are run on instructions from the central government in order to make the central power more effective. In France there are, for example "les departements", and in Sweden we have our "län". These are important components of typical hierarchies.

In many governing hierarchies levels are defined in terms of areas or geographical districts. There are, however, also other kinds of hierarchies in government, which are not necessarily so defined and which may be used for slightly different purposes. In public administration, in particular, there are two purposes that are of interest in this context, namely the quest for general overview, and the quest for uniformity.

As an example of the first purpose we may consider the energy policy of a country. Here the legislature may decide about the broad lines and allocate specified amounts of money to solar energy, nuclear energy, energy saving, and so forth, and these sums may be handed out to authorities that are specialized in these various fields. The authorities, in turn, may distribute the money to their own departments, which are further specialized, and these departments, perhaps, will finally send the money to the intended receivers - for example people engaged in various kinds of relevant energy research. A system like this is obviously hierarchical since all decisions are intended to serve the national energy policy. The system also gives the decision-makers, and in particular those at the top of the hierarchy, a good overview of what is happening. The other side of the coin, however, is an obvious element of central planning. In many areas, such as for example energy policy, the merits of the whole arrangement can thus be disputed. Still, hierarchies of this particular kind, serving the purpose of overview, are common in public administrations.

As an example of the second purpose we can consider a national judicial system. In such a system there are often courts at several levels and, if so, it may be possible to appeal against the ruling of a particular court at a court at the next level upwards. Even if this system is not hierarchical in the sense that higher courts give orders to lower courts, it is so in the very similar sense that higher courts may, first, change individual verdicts of lower courts and, second, thereby sometimes also create precedents that constrain future action of lower level courts. One rationale for this system is its likelihood to enhance the quality of the verdicts, but the uniformity that it fosters is usually also, since it means equality before the law, considered a value by itself.

The size of the hierarchies used in governing do not usually have natural size limits as, for instance, the hierarchy created for building a highway. The latter's size was determined by the task to be performed. A public hierarchy does not have a natural size limit like that. Consider, for instance, the quest for overview. It could always be argued that a better overview could be obtained by putting more policy areas under the same hat - environmental policy could for example be linked to energy policy with fewer and bigger hierarchies as a result. Thus, bureaucratic free enterprise may be a permanent threat, and a quite likely phenomenon, in public hierarchies used for governing.

Federations 3

A federation, like a hierarchy, has several levels, although, as in figure 2, the number is usually limited to two - a national or federal level, and a state level. The relationship between the levels differs, however, completely from that in a hierarchy. In a federation there is no such thing as a common, all-pervading goal. On the contrary, the very idea is that the entities at the various levels should be free, within specified limits, to decide about their own goals and activities. An example may be illustrative. Imagine that there are a number of adjacent but separate nations, each one characterized by its own history, its own culture, and its own language. Imagine furthermore that these nations are threatened by a common enemy, which is much bigger and stronger than anyone of them individually. If so, it seems a good idea for the nations to engage in some kind of common defense. One possibility is to create an ordinary defense alliance. That kind of solution may be enough, although it has some shortcomings. Since, in such an alliance, there is no common central authority with power to tax the member-nations for building an efficient military organization, there will be incentives to free-riding and risks of failure. This may be avoided by going one step further and creating a common authority with power to tax and to act in the military area. At the same time, however, the member nations, because of their different characters, have all reasons to keep their remaining powers intact. Thus, the result is a federation in which military decisions are taken at the central, federal level, and all other decisions at the level of the former nations, or states.

Figure 2

Riker gives the following definition of a federation (Riker, 1964, p 11): "A constitution is federal if i. two levels of government rule the same land and people, ii. each level has at least one area of action in which it is autonomous, and iii. there is some guarantee (even though merely a statement in the constitution) of the autonomy of each government in its own sphere." The opposite of a federal government is a unitary government.4

A few comments on the meaning and implications of the autonomy concept may be added. First, since Riker talks about two levels of government ruling the people, I interpret him as saying that the autonomy, at both levels, includes the power to legislate (within the specified limit). Second, since the autonomy of the two levels is crucial in a federation, and since the borderline between the two levels has to be clear, a constitutional court seems to be a natural part of a federation. This position is in fact quite common. Bealey (Bealey, 1999, the entry "federal government"), for instance, maintains that "a constitutional court must act as an arbitrator between the different authorities."

Federalism is thus associated with a legal structure. In particular, there are laws that divide the power between the levels, and there is also, in all likelihood, a court for settling conflicts between the levels. Each level has also a capacity to legislate within specified limits. When these legal conditions are fulfilled we may talk about a real federation. Even without these legal prerequisites a leveled structure may, however, have great similarities with a federation in the real sense. There may be a division of power between the levels that holds in practice, even if not specified in law. And the lower level may have considerable independence without having, in particular, the capacity to legislate. In such cases we may speak about quasi-federalism.

Obviously federations may differ with respect to the amounts of power, generally speaking, which are allocated to the federal level, and to the state level, respectively. If most power is allocated to the federal level we speak about a centralized federation. If, on the contrary, the power basically rests with the states we are dealing with a peripheralised federation. The terminology is taken from Riker (Riker, 1964, p 5f).

There are many federal nations in the real world. In 1964 Riker (Riker, 1964, p 1) made the following lists for different parts of the world:

Since he was concerned only with federalism as such, Riker had both democracies and dictatorships in his list.5 In 1999 Arend Lijphart, who was concerned only with democracies, made another list. He classified 36 nations in the world as democracies. Among these he labeled some as federal and some as semi-federal in this way (Lijphart, 1999, p 189): Lijphart's definitions of federalism and semi-federalism, it should be noted, do not exactly correspond to the notions of real federalism and quasi-federalism above. They are, to put it briefly, somewhat less legalistic and somewhat more sociological.

Here I have used the term "state" for the entities at the lower level, but in reality the vocabulary is shifting on this point. In the United States, India, Australia and Venezuela they are, in fact, called states, but in Canada they are called provinces, in Germany and Austria Länder, in Switzerland cantons, and in Belgium regions (according to Lijphart, 1999, p 187).

Mixtures of Hierarchical and Federal Elements

Hierarchical and federal structures sometimes appear side by side, and sometimes almost imperceptibly so. Consider for example the local communes in Sweden. They have two sectors, or two areas of decision-making, which are usually called the bounded sector, and the free sector, respectively. Within the bounded sector the communes essentially do things that they are required by the state to do - they are, for example, running schools as prescribed by the national laws. In this area the communes are local extensions of the central national administration, and thus operating within a hierarchical structure. In the free sector the situation is completely different. There the communes undertake any activities they want, subject only to the obvious restriction that they have to remain within the boundaries of relevant laws. They may, for instance, build various kinds of sporting grounds, or they may keep away from such activities completely. Within the free sector the communes are thus operating within a federal structure, or, more precisely, within a quasi-federal structure.

Having mentioned the coexistence of hierarchical and federal elements in the Swedish municipalities it should also be mentioned that not even federal nations are always as federal as they may appear. In Argentina, Brazil and India, for example, there are constitutional rules about "federal intervention to govern states that fail to perform satisfactorily (when of course, the federal government is judge of what is satisfactory performance)" (Riker, 1964, p 120). The implication, of course, is that the states are not really free, and that, to that extent, we are in effect dealing with a hierarchy rather than a federation.

2. The Unitary System

The Border

In order to be able to discuss the properties of two-level federal systems it seems expedient to consider one-level systems, i e unitary systems, first. An understanding of unitary systems is, I think, a prerequisite for a good understanding of multilevel systems. In the discussion about unitary systems I will consider both the mechanisms behind the outer limits, or the borderlines, of such systems, and the problems of governing within those limits. These two kinds of problems may to a large extent be treated separately, but nevertheless they are also closely related to each other.

As regards the discussion of the outer limits I will start by quoting Riker. "National decision making", he argues (Riker, 1964, p 146), "is in the abstract more efficient than local decision making on every issue. That is, ... , it is cheaper on any subject of legislation to have a uniform rule, ... , than not to have uniformity." This is an argument about advantages of scale. Some things benefit from being done on a large scale rather than on a small scale. Among these things, according to Riker, are legal rules and legislation.

This argument by Riker may be systematized and generalized. For that purpose I will use the "map" in figure 3, which represents two different cultural-linguistic areas with a common border. The area A, which is divided into a number of sub-areas, is a bit larger than B, and we may also assume the same about its population. Let us at first confine the discussion to A and start by considering traffic law, for instance about traffic lights, which is discussed by Riker. Two possible arrangements may be compared. In the first one each of the nine sub-areas decide about their own rules, whereas in the other one the rules are common to the whole A area. Which are the advantages and disadvantages of these two arrangements that we may call the local arrangement and the encompassing arrangement respectively?

Figure 3

This question is important but even so it is not clear how it should be answered. One method, which has been used in slightly different variations by Riker as well as by Buchanan and Tullock, for treating this kind of problem, is to identify and take into account different kinds of relevant costs. I will follow the pioneers on this track even if my method, in turn, is a slightly new variation: the list of costs is not exactly the same.

I will start with the decision-making costs, which is a concept introduced by Buchanan and Tullock. These costs, basically a kind of transaction costs, are the costs associated with arguing, building coalitions, and so forth, in order to achieve a decisive constellation.6 If we assume that both the local and the encompassing arrangement utilize some kind of democratic decision-making, with a specified type of representation and a specified decision rule, then the total decision-making costs, per capita, will be lowest in the first case. Thus, from this point of view, the local arrangement comes out better than the encompassing one. But there are more aspects to the problem than this.

Another kind of costs, discussed by Riker, is the costs of law enforcement. These costs, according to Riker, are higher in the first case. "If a red light meant different things in different localities", he says, "it is highly likely that the costs of enforcement of road rules would increase enormously." Riker is certainly right here and this aspect thus favors the encompassing arrangement rather than the local one. But even so it is important to make clear exactly why Riker is right. Law enforcement is necessary because people sometimes tend to break the law. The reasons for not following the rules may, however, vary considerably. People may break the law since they derive some kind of advantages from it, or they may break the law by mistake, due to ignorance. Furthermore, in a heterogeneous cultural-linguistic environment many people may break laws simply because they do not acknowledge them. In the example here, since we are dealing with a homogenous cultural-linguistic environment, the first two reasons are the most likely. It is also likely that the distinction between the local and the encompassing arrangements is irrelevant to breaks on purpose, whereas it is highly relevant to breaks by mistake. The costs of law enforcement in a homogenous environment are therefore related to another factor that Riker also mentions, namely people's efforts to learn the law.

The costs for those affected of learning the laws are obviously smaller the fewer the laws are. If we consider, for instance, traffic rules and take for granted that people, to a large extent, drive around in the whole linguistic-cultural area, and not just in their own local sub-area, the local arrangement means that they will have to learn nine laws, whereas the learning of one law is enough in the encompassing arrangement. As there are many people involved, these costs are likely to be considerable, and this factor thus strongly favors the encompassing arrangement. It should be added that those who have to learn a law, since they are affected by it, are not only ordinary people but lawyers as well. This may be more obvious if we turn from traffic to trade, for example. In this case not only the traders but also, and especially, their legal advisors have to be familiar with the legal preconditions.

The factors so far brought into the discussion thus indicate that the encompassing arrangement is better than the local one. There is, however, in the traffic law example, one important prerequisite for this conclusion, namely that technology permits large numbers of people to leave their local area and travel freely in a wider region. The conclusion favoring the encompassing arrangement is thus technology-dependent, and this result may perhaps be generalized. If technology develops, and makes activities of various kinds possible over wider and wider areas, the case for more and more encompassing laws becomes concomitantly stronger.

This being so, it is straightforward to continue by asking about the prospects for laws that are still more encompassing, and whose application thus transcends the borders of the linguistic-cultural areas. Consider, for instance, a suggested law for an area consisting of all of A plus all of B. In such a case the costs of learning will obviously decrease, since it is enough for those affected to learn one law rather than two. The costs of the decision-making may however increase enormously, especially if the two cultures are different enough, and the same is true for the so-called external costs, in Buchanan-Tullock's sense, which are the inconveniences from a decision hitting those who do not support it. Imagine, for instance, that the tolerance for alcohol differs considerably in the two cultures, and that this is relevant in relation to traffic and driving as well as in other ways. If so, and if a majority rule is used, the external costs may become very high. If, for instance, a majority from A overrules a minority from B, the inconveniences for the latter may be considerable. In this case, since the law is not generally accepted, the costs of enforcement are also likely to be considerable. If, on the contrary, some more demanding decision rule, for example some qualified majority rule, is used, the decision-making costs may raise to very high levels since it may be very difficult to find a compromise, perhaps including a scheme for side-payments, which gets the necessary support. Consequently it may not pay to extend the law across cultural-linguistic borders.

For many laws the homogenous cultural-linguistic area is thus a natural area of application. The explanation, as we have seen, is that the expansion of the application to the cultural-linguistic borders is easy and profitable, whereas further expansion may meet considerable obstacles. Having said this we have also provided an answer to a question that is almost never asked, but nevertheless deserves a serious treatment, namely the question of the nature of the nation. At least on the surface, this question has similarities with Coase's seemingly innocuous question about the nature of the firm. As for nations we may ask why they are there at all, and why they look the way they do. Why do their borders usually, and at least to a large extent, coincide with the borders of cultural-linguistic areas? Why does not nation-building follow other principles, which disregard the cultural-linguistic borders? What exactly makes these particular borders so important? A likely answer, as we have seen, is that the cultural linguistic area is often optimal to legislation. Furthermore, we may now assert that the state is the entity within the nation that provides the legislation. This, thus, is the basic rationale for the state. Other common notions of the state, for instance that its raison d'être is the provision of public goods, are at best misleading. True, laws are public goods, but they are a very particular kind of such goods, and there are many other kinds of public goods. So the rationale for the state, within the nation, is the provision of laws and law enforcement, not the provision of public goods in general.

The importance of this result, in this context, is that it indicates an absolute scale for the allocation of levels in a multilevel system. After this we can do more than just order the different levels relatively, as being below or above each other. The national level serves as a base level, or a level of departure, when considering other levels, upwards or downwards.

Legislation and Implementation by Hierarchy

After having considered hypothetical mechanisms behind the delimitation of nations, I will now consider the problem of governing a nation. This, of course, is a giant complex, and I will just concentrate on a few aspects of relevance in relation to federalism.

A first point is that not only the top-level position of legislation, but the bottom-level position of the practical implementation of various tasks as well, are supported by some general principles. To illustrate this, I will use an example inspired by Buchanan and Tullock (Buchanan & Tullock, 1962, pp 111 f). Suppose that a school is to be built and run for the people living in a certain area. Suppose, furthermore, that the school is to be public rather than private, and that it shall be financed collectively rather than by vouchers (a case considered below). If so, the decisions about the school - for example about its location, about the building, about which teachers to employ, and so forth - are to be taken collectively. So the problem is: which collective is the appropriate one?

Let us assume that there is a limited group of people who start thinking about the school since they want it for their children, and that these people are indicated by the hatched innermost part of figure 4. These people obviously should be members of the decision-making group, or at least they should be represented in it. Now, since the school is assumed to be run as a collective good there are, in all likelihood, also other people who may benefit from it, and who are thus likely to use it for their children when it is built and ready. It therefore seems to be a good idea that these people, who are marked by the dotted area in the figure, are brought into the decision-making group as well. More specifically, and according to the argument by Buchanan and Tullock, two things happen when the decision-making group is expanded. One is that the decision-making costs per individual increase, since it requires more efforts, per individual, to reach a decision in a big group than in a small one. On the other hand however the risks of harmful, or bad decisions, or in other words the external costs, become smaller the more of those affected by the decisions take part in them. This effect may very well compensate for the former one and thus, to a point, justify the expansion of the decision-making group. After that point however, since the people who still are affected by the school either are too few, or too little affected, the increased decision-making costs which follows from a further expansion of the decision-making group, for instance to the outer borderline in the figure, will not be fully compensated for by better decisions. The collective activity should thus, according to Buchanan and Tullock, be organized "in the smallest units consistent with the externality that the collectivization is designed to eliminate."

Figure 4

This reasoning, which uses the same general kind of arguments as the discussion about laws in the preceding section, thus leads to the conclusion that a lot of practical, administrative and public activities can, and should, be organized on a small scale locally. In addition to schools this may be valid for roads, harbors, bridges, parks, museums, and so forth.

Now I have come to a position where it is possible to put Riker's argument about the benefits of legislating on a large scale, and Buchanan-Tullock's argument about the benefits of handling practical administrative tasks on a small scale, into a common framework. There is, in fact, a passage by Milton Friedman where his main concern is to argue in favor of school vouchers, but where, incidentally, he also happens to supply exactly the kind of framework I am interested in here. "I distinguished", he recalls in his memoirs (Friedman, 1998, p 347), "among three levels of government involvement in schooling: requiring schooling, financing schooling, and administering schools. I argued that a case could be made for both the imposition of a minimum required level of schooling and the financing of this schooling by the state. However, I went on, a third step, namely the actual administration of educational institutions by the government ... is difficult to justify on any grounds. Governments could require a minimum level of schooling financed by giving parents vouchers redeemable for a specified maximum sum per child per year if spent on 'approved' educational services. Parents would then be free to spend this sum and any additional sum they themselves provided on purchasing educational services." This argument for a voucher system in education is interesting here for three reasons.

First, Friedman explicitly describes education as a complex undertaking involving legislation, financing and the running of individual schools, and argues that these different tasks belong to different levels. This is in sharp contrast to simplistic notions such that, for instance, education is a local government service. Public tasks, as well as private ones for that matter, are often complex in this sense. Friedman obviously has a hierarchy in mind, although he does not mention that particular word, and some basic characteristics of that hierarchy are noteworthy. Thus, legislation takes place at the top level, which is in accordance with the principles formulated in the preceding section. This is very much as the general staff at the top of a military hierarchy formulating the general strategic and tactical principles for an impending operation. Furthermore, the practical implementation of the principles in the law, for example the running of individual schools, takes place at lower levels, which is also in accordance with the principles presented above. In the military analogy this corresponds to the individual platoons, and even soldiers, performing their practical assignments, down to earth tasks, at the low levels of their hierarchy.

Second, Friedman's argument makes an explicit case for private alternatives, even as hierarchical elements, since the schools that receive the vouchers may very well be private enterprises. This is something that should be kept in mind whenever tasks to various actors are assigned.

Third, when Friedman says that parents should be free to spend the voucher sum, and any additional sum, on educational services, he introduces a further and different kind of private activity. The voucher sum belongs to the hierarchical system, whereas the additional sum, which is not compulsive, is a result of an initiative from the parents. Initiatives of the state, as represented by the hierarchy, are thus mixed with private initiatives from the customers-consumers.

This example from Friedman illustrates a kind of hierarchy that is omnipresent in public affairs. At the upper level legislation takes place, at the lower levels implementation. The activities at the lower levels, although always publicly financed, may otherwise be either public or private. Furthermore, the introduction of private consumer initiatives may be tolerated by the hierarchy, and may even be important. This kind of hierarchies is, I think, not only a common but also a very useful, and often even necessary, governing device.

The existence of hierarchies of this type does however not imply that all local, public, and administrative undertakings are components of such hierarchies. There are, obviously, also a lot of activities which are run as independent enterprises on their own merits. Buchanan-Tullock's principle regarding the benefits of small units may thus sometimes concern activities that belong to hierarchies, sometimes independent activities.

The Principle of Subsidiarity

Buchanan-Tullock's principle is, in fact, closely related to, or rather a variety of, a principle saying that "the responsibility for functions should be allocated, as far as practicable, to the lowest levels of society". This frequently invoked principle, usually called the principle of subsidiarity, is in fact very old and attributed to Thomas Aquinas (1225-1274). He had the problem of protecting the Catholic Church from the infringements of the state, and to that end he proposed that all activities of the church should be performed at the longest possible distance from the threatening state, and thus at the lowest possible level. (Bealey, 1999, the entrance "subsidiarity"). But even if the principle dates back to the thirteenth century it is, as I have indicated, still going strong. It is, for instance, prominent in present thinking about European integration and, in fact, formally inscribed into the Maastricht Treaty (Oates, 1999, p 1122).

Here it is important to note that this principle is compatible with federal as well as hierarchical ideas. Thomas obviously thought along federal lines. He wished the church to enjoy real federal freedom in complete independence of the state. Even in an hierarchy it is however possible to argue that decisions should be placed at the lowest possible levels, since this is a way of placing the decisions where the information is, and thus of avoiding top-heaviness.

Risks for Bad Government

Governing, of course, is difficult and risky, and may go astray. In particular there are the risks that those in power satisfy their own wants at the expense of those of the citizens. One kind of threat is that hierarchies, although to a large extent necessary, become too large, and too top-heavy. There are also other ways in which the governing power may become too concentrated, too strong, too unrestrained, and too uncontrolled. Several imaginable means have been suggested for correcting such failures, among them the introduction of a certain amount of real federalism. Thus, the idea is that some of the problems of the unitary state may be alleviated by federalism. This idea will be discussed in chapter 4. Before that, however, I will discuss some general topics related to federations.

3. Federalism - Some General Aspects

Mistaking Hierarchies for Federations

We have seen that hierarchies and federations, although they are similar in being multilevel systems, are also strikingly different. In hierarchies lower actors get orders from higher actors, whereas, in federations, lower actors enjoy freedom, upheld by law, from intervention by higher actors. Considering this fundamental difference, it is interesting to note that the two kinds of systems nevertheless are often mixed up, even by experts.

Consider, for instance, the following passage from Riker in his work about federalism, under the headline "Transportation and communication" (Riker, 1964, p 68): "The usual rule (not always followed, of course) for the division of labor on this function among the several levels of government seems to have been: The smallest government that rules both terminals of the proposed line of transport or communication is the one to promote and supervise it. The operative effect of this rule is that initially the states were the main promoters, whereas in the last century or so the United States has become the main one. Even in the beginning, however, the central government had quite a bit to do and today the states have quite a bit to do. At no point is there an exclusion of one kind of government from the function. Rather, the function is fully shared. It is just that the recognized rule of division gave the states the preponderance of the task in the beginning and now gives the preponderance to the nation." What Riker describes here, in spite of his intentions, is not a federal system at all. Rather, and obviously, he talks about a hierarchy in exactly the sense defined and discussed above.

In another context Riker writes about Germany (Riker, 1964, p 123), that "... the federal government does not directly administer programs in many of the areas of action about which it legislates, ...". Here he is explicitly dealing with federal institutions and with degrees of centralization and peripheralisation. Still, he obviously describes a hierarchy: the federal government legislates about programs that are administered by the "Länder". The hierarchy, in fact, fits very well into the pattern described in the section "Legislation and implementation by hierarchy" in the preceding chapter.

Or consider Buchanan when he writes (Buchanan, 1999 b, p 311) that: " 'Federalism,' 'devolution,' 'decentralization,' 'subsidiarity,' 'denationalism,' 'privatisation' - these are political terms of the 1990s. The current relevance of these terms stems from the wide spread acknowledgment that, either directly or indirectly, regulatory authority has been too much concentrated in central governmental units, with the result that they have become omnipresent. The argument from organizational efficiency alone dictates institutional reform in the direction of decentralization." To me, at least, it seems that Buchanan here, in spite of what he says himself, is concerned with hierarchies rather than with federations. He purports to talk about "federalism", etc., but goes on with arguments about a too concentrated regulatory authority, and about organizational efficiency. What he criticizes here, I think, is top-heavy hierarchies that would benefit from more delegation downwards. The concepts of regulatory authority and organizational efficiency belong to hierarchical rather than to federal thinking. Hierarchies really exist in order to enhance efficiency. The basic purpose of federations, on the contrary, is the delimitation of areas of independence or freedom.

In the following I will give some further examples of hierarchies mistaken for federations.

Reasons for Two-Level Systems

But even if hierarchies are sometimes mistaken for federations there are real federations as well. Whenever we are dealing with such a system there is a problem about the reason for the two levels of decision-making. Why are the constituent states not, in all respects, acting as separate entities? Or, alternatively, why do they not merge all their activities, and thus form a common unitary state? Why do they find it important to keep their individuality in some respects? In the example in the first chapter the defense problem provided the answer to the first question. The individual states were not able to defend themselves. Had it not been for that problem the states would not have formed a federation. They would have remained as separate unitary nations in all respects. As for the second question the answer is in terms of the different cultures and histories of the constituent states. Had it not been for those differences the states could as well, together, form a single, big, unitary nation.

These points certainly go a long way towards explaining the two levels, but still there are some unanswered questions. The most important one is why the military problem requires a fusion of the constituent states into a federation. Why is an ordinary defense alliance not enough? The answer given in chapter one is that the threat of free-riding is countered much more effectively in a federation than in an alliance, and this is certainly true. Still, even this answer is not complete.

The problem is that the group of nations forming the federation is quite small, whereas free-riding, according to Mancur Olson's logic of collective action (Olson, 1965), is usually much more of a problem in large groups than in small ones. Therefore, if free-riding is to become a real threat, there must be something which makes it especially tempting even in a small group of nations. The fact that military expenses are usually extremely high may account for that.

Two factors thus basically explain the defense federation. First, the constituent states are too small for defending themselves, and therefore they have to cooperate. Second, since defense expenditures usually are very high, a federation is a better solution than a defense alliance. Now, it is of course interesting to ask about the possibilities to generalize this result. Are there, besides military affairs, any other policy areas which call for federations for the same reasons? Are the conditions of scale, and of costs, fulfilled in any other policy areas?

Consider for instance environmental policy, and suppose that global warming due to carbon dioxide emissions is a real threat. If so, cooperation between nations is a necessity. Furthermore, since it may be very costly to avoid emitting carbon dioxide, there may be a very strong incentive for free-riding, and thus a justification of large federations, and ultimately a global one. The problem, which we shall not deal with here, is that the threat of global warming is sometimes disputed.7 But even if this threat is a real one I fail to see a lot of similar cases. Usually, and introducing a normative argument, I think, international cooperation could, and should, take place in ordinary organizations, such as for example Interpol, rather than in federations.

Problems with Two-Level Systems

Having thus discussed some rationales for two-level systems it is appropriate to consider some problems as well. In the literature some such problems are discussed and they are usually related to the interaction between the different parts of a federation, for instance the interaction between the states, or between the two levels.

Riker warns that legislation in the states, in a federation, often is tantamount to minority rule. One example is about minimum wages (Riker, 1964, p 146). He assumes that one state (in the US) has a lower minimum-wage level than all the other states have, and want to have. If so, capital that yields more in a low minimum-level environment will migrate to the low level state, and all other states would suffer. The only way to counter this, apart from national decision-making, is to follow the single state and decrease, in all other states, the minimum wage level. This means that the single state with the initially low minimum wage sets the national norm, and thus the minority rules.

Another example is about divorces (Riker, 1964, p 147). There are according to Riker a few states in the US, in particular Nevada, which grant divorce "on the most trivial grounds and after only the briefest periods of residence". He argues that these rules have been motivated mainly by a wish to attract tourists, and that they have "the effect of imposing easy divorce on all jurisdictions." ... Therefore "(t)he external costs of the existence of what most citizens of most states regard as low moral standards are thus very high for the probably overwhelming majority that opposes easy divorce." Again, says Riker, a policy is imposed on the majority by a minority.

Riker's sharpest criticism of US federalism, and its tendency towards minority rule, is however related to the race issue. "The plea for states' rights or for maintaining the guarantees of federalism", he contends (Riker, 1964, p 142), "is simply a hypocritical plea for the special privilege to disregard the national majority and, of course, to permit one minority, segregationist Southern whites, to tyrannize over another minority, the Southern Negroes." Then (Riker, 1964, p 144) he concludes that: "For the last century approximately 10 per cent of the people have been denied civil rights. Those so denied have been a kind of lumpenproletariat and hence a drain on the whole society."

Surely some aspects of Riker's examples may be discussed. Thus it is obvious that his worries are not limited to the minority rule as such, but also include the particular results illustrated by the examples. In this latter respect there is, however, clearly room for other opinions. But even so the minority-favoring mechanism may clearly be a reality, and something to be concerned about.

Other kinds of ominous interaction between the parts of a federation are related to financial flows. Assume, for instance, that one of the constituent states tries to attract an enterprise by offering particularly favorable tax rules, and as a consequence gets a public finance deficit. If so the federal government, for one reason or another, may step in and cover the deficit. In the long run that kind of deficit-creation, and covering, may obviously ruin the whole federation. The mechanism described, which is related to the dynamics between the components of a federation, is an example of what is sometimes called "race to the bottom".

Some further problems with two-level systems are discussed in the section "The framework for the competition" in the next chapter.

Fusion, Fission and Equilibrium

The two levels of a federation are never, as it seems, created simultaneously. Either the upper level, or the lower level, comes into existence after the prior, original level. In the former case the federation is a result of fusion, in the latter a result of fission. These two cases are illustrated by the examples 1 a and 1 b, respectively, in figure 5.

Riker has studied empirically the mechanisms by which the present federations in the world were created (the ones mentioned in the list taken from Riker in the first chapter above), and his results are therefore of great interest here. When formulating his hypothesis Riker conceives of a federation as a result of a bargain according to which some actors create power at an emerging federal level, and others give up power at the state level. The motives of these two groups are described by Riker as follows (Riker, 1964, p 12):

1. The politicians who offer the bargain desire to expand their territorial control, usually either to meet an external military or diplomatic threat or to prepare for military or diplomatic aggression and aggrandizement. But, though they desire to expand, they are not able to do so by conquest, because of either military incapacity or ideological distaste. Hence, if they are to satisfy the desire to expand, they must offer concessions to the rulers of the constituent units, which is the essence of the federal bargain. The predisposition for those who offer the bargain is, then, that federalism is the only feasible means to accomplish a desired expansion without the use of force.

2. The politicians who accept the bargain, giving up some independence for the sake of the union, are willing to do so because of some external military-diplomatic threat or opportunity. Either they desire protection from an external threat or they desire to participate in the potential aggression of the federation. And furthermore the desire for either protection or participation outweighs any desire they may have for independence. The predisposition is the cognizance of the pressing need for the military strength or diplomatic maneuverability that comes with a larger and presumably stronger government.

Riker then hypothesizes that these motives are always present when a federation is created, and that the two points therefore can be regarded as necessary conditions for a federation. After the empirical investigation he finds that his hypothesis is confirmed. "On the basis of the evidence here set forth, he states (Riker, 1964, p 48), "we may conclude that the hypothesis is confirmed that the military and expansion conditions are necessary to the occurrence of federalism."

Being more detailed, Riker (Riker, 1964, pp xii, 1, 8, 10, 34) furthermore maintains that all existing federations apart from medieval Switzerland are creations of the 19th and 20th century and thus modern. They are inspired by the US constitution. The US federalism was centralized rather than peripheralised and this was, according to Riker, the main reason for its appeal and influence. Central federalism was invented in Philadelphia in 1787 and was a crucial event in the history of federalism.

Figure 5

Switzerland, which today is a unique federation (in being peripheralised), was originally not at all so. There were many of its kind, but the others were conquered and disappeared. Their military performance was inadequate since their federalism was peripheralised rather than centralized - in the same way as for the Greek city-state-federations long before. Or, in Riker's words (Riker, 1964, p 8): "In northern Italy and southern Germany medieval cities formed military federations to resist the encroachments of nascent nation states, all of which were essentially imperial in nature. These Italian and Suabian federalisms went through exactly the same cycle as the Greek federalism had done 1,500 years earlier and for exactly the same reasons. Only one of them survived into the modern world: the Swiss confederation and it survived not because of its constitutional form but because of its unique geographic advantages for defensive military operations."8

Anyway, Riker's main result means that all existing federations were created for some kind of military or diplomatic reasons, and that all came about by fusion. The 1 a pattern is thus common. On the contrary there are no historical examples, at least among the present federations, of the 1 b type.

The frequent 1 a process, as well as the infrequent 1 b one, means that stable federations are created. Earlier in this chapter I mentioned the problem of the raison d'être of the two levels, and I indicated that stable federal multilevel systems are quite remarkable, and nothing to be taken for granted. One may say that there is a balance, or equilibrium, between the levels, which calls for an explanation. There are, in the literature, some examples of this kind of thinking. Thus Buchanan explicitly talks about equilibrium (Buchanan, 1999 a, p 182). "I like", he says, "to think of a federal structure in the following way, and I think it is applicable in the European context: Think of a satellite in orbit. A satellite in orbit is, of course, being maintained by the centripetal and centrifugal forces in equilibrium. So you can sort of think of federalism as a satellite being pulled always by two separate forces in opposite directions - namely, a force toward monolithic centralized authority and forces in the other direction toward a set of autonomous separate units. So federalism is a kind of equilibrium. It is shaky to maintain that equilibrium, but I think in the European context it may be the future likelihood or prospect."

A federal equilibrium, since it is a delicate thing, is by no means an assured result. If a process in the direction of federalism starts, whether upwards or downwards, it is not certain, and it may in fact not even be intended, that it should halt in a state with two coexisting levels. The central power may progressively impose its will on the lower level and thus reduce the original freedom there. In that way a federal system may gradually, and perhaps unnoticeably, turn into a hierarchical system. This is the 2 a pattern in the figure, which represents a complete fusion or amalgamation of the original constituent parts. It is, however, also possible that the upper level is challenged, and loses. If so we have a case of the 2 b pattern, which represents a complete fracturing or fission of a single original entity.

Probably there are many historical examples of the 2 a pattern. What I am thinking of is the unification of the Netherlands, Italy and so forth. The 2 b process may perhaps be illustrated by the disintegration presently taking place in the different parts of the prior communist world.

In the patterns 3 a and b of the figure a process towards federalism, for one reason or another, finally proves untenable, and turns back. Here, the European unification may be used as an illustration. Mostly, I think, we tend to consider it as a process of type 1 a. We thus tend to think about the EU as an emerging, viable and stable federation with two levels. That may however be premature. The alternative 3 a, for instance, may still be open. This means that the unification fails, and that the nation-states fully reappear. In addition to that, and going back to the earlier patterns, 2 a may also be open. If this alternative materializes, the European federalism turns into a European hierarchy. Buchanan does not predict this alternative, but he has described it vividly while criticizing the Maastricht treaty and the dirigistes from Brussels (Buchanan, 1999 a, pp 181 f). '"The danger", he argues, "is that the Brussels bureaucracy will dominate and that you will have excessive harmonization and regularization, which would mean a less prosperous Europe."

4. Competition in Law

Competition in Policy

Market competition is usually considered beneficial. According to this view, the competition between firms for customers is a powerful mechanism for the enhancement of general welfare. In this process the firms are striving to maximize their profits, and the customers to maximize the satisfaction derived from their resources. This idea of competition is sometimes transferred to the realm of polities and citizens. It is imagined that polities, in analogy with companies, and by varying their policies, can compete for citizens, who are thus the analogues to the customers. It is furthermore imagined that the good polities, i e the polities in which the citizens lead relatively decent and happy lives, will somehow survive and prosper in the process, whereas the bad polities, if they don't change their policies, will somehow get marginalized. A basic process involved is migration. People leave the bad polities and move into the good ones.

Ideas like these have been expressed by several writers. Buchanan and Tullock argued for instance in The Calculus of Consent (Buchanan & Tullock, 1962, p 114), after having made the case for the smallest possible units of collective activity (as accounted for in chapter 2 above), that "If the organization of collective activity can be effectively decentralized, this decentralization provides one means of introducing market like alternatives into the political process. ... In concrete terms, this suggests that the individual will not be forced to suffer unduly large and continuing capital losses from adverse collective decisions when he can move freely to other units, nor will he find it advantageous to invest too much time and effort in persuading his stubborn fellow citizens to agree with him." After this Buchanan has remained faithful to the idea throughout the years. Thus he wrote very recently (Buchanan, 1999 a, p 179) that "If there is an exit option, if there is a chance to leave, this necessarily imposes discipline on those who would exploit you through a political structure or a bureaucratic structure."

Furthermore, Buchanan has explicitly associated this idea of competition with federalism. "For me", he thus writes (Buchanan, 1999 b, p 312), "the ideal political structure, especially in large polities, is one described by competitive federalism: the central government remains strong but severely limited in its range of powers, and the separate but integrated lower-level units compete with one another in providing for most of the collectively supplied services as demanded by the citizenry." Then he goes on saying that "(t)he most exciting potential application of genuinely federalist principles seemed to me to be present in the European thrust toward effective political and economic integration. In 1990 I presented a paper in Paris, 'Europe's Constitutional Opportunity,' in which I outlined what seemed to be a once-in-history chance for Europe to establish a strong, and limited, central authority with competing national units and open internal markets." Things did, however, go other ways and later on (Buchanan, 1999 a, p 181 f) he deplores that "... Europe did miss out on a wonderful opportunity to introduce a really good monetary structure. It seems to me that Europe in the early nineties had the chance of adopting a proposal for competitive currencies. All that would have been needed would have been something in the treaties that would have allowed any citizen of any country in Europe to make contracts, including paying taxes, in any money that is issued by any bank in the structure. In that context, the separate central banks could still operate, and you could still have the local currencies, but in fact they would be in competition, so no central bank would dare inflate the currency because it would essentially lose out, and therefore you would have discipline imposed by competition on the separate banks. At the same time, people could have still retained their traditional currencies. Instead, Maastricht tore up that kind of possibility and they moved toward the Euro. I think Europe missed out on a wonderful opportunity."

In his recent proposal that the generality principle should be applied to politics Buchanan has also, together with Congleton, argued in favor of competition in policy, and for federations as fitting frameworks for harboring such competition. The competition is, it should be noted, a "long-term competition for residents" (Buchanan & Congleton, 1998, p 141). These arguments are, however, less general than the ones quoted above, since they presume that the generality principle is in fact applied. At least for the moment this is a very special, and unrealistic, condition. I will come back to this point later on.

Another scholar who, in a way, has written about competition in policy is William Niskanen. According to him (Niskanen, 1971, p 19) "Even the most autocratic government would have to be approved by every resident if the costs of moving were zero. Effective popular control of government is more important, the greater the continuous territorial sovereignty of the government and the greater the restrictions on moving across the borders of the state, both of which increase the economic and psychic costs of moving. Unfortunately, the difficulties of popular control of government also increase with the size of the state. The superiority of ... local governments in providing certain services is due more to the lower cost of ... moving than to greater individual control - which, except in the smallest organizations, is still negligible. A purely autocratic government of a small area must be more responsive to each resident, if there are no coercive controls on moving, than the best conceivable democratic government of a large state." Niskanen thus argues that local government, for some public services, tends to be superior to national government. The reason is the low costs of moving from one governmental area to another, both due to the relative smallness of local government areas, and to the fact that the passing of borders between local governmental areas usually require much less formalities than the passing of national borders.

Another contribution to this field of thinking is Charles Tiebout's (Tiebout, 1956). His version of the ideas involved is less sketchy, and more elaborated, or formalized, than those of most other researchers. Since he specifies explicitly the premises of the results he deduces, he is in fact presenting a real theory or model and not just a mere suggestion.

Tiebout started with the well-known problem of determining the quantities of collective goods to be provided by public authorities. The conventional wisdom is that this problem has no exact solution, since, as Tiebout argues (Tiebout, 1956, p 417), "... there is no mechanism to force the consumer-voter to state his true preferences; in fact, the 'rational' consumer will understate his preferences and hope to enjoy the goods while avoiding tax." Tiebout's treatment of this problem starts with the contention that some public goods such as "police and fire protection, education, hospitals, and courts" may be provided locally. These kinds of goods may thus be provided in different combinations by different local governments, and hence the individual consumer voter may also move to "the community that satisfies his preference pattern" (Tiebout, 1956, p 420). "Moving or failing to move", says Tiebout (Tiebout, 1956, p 420), "replaces the usual market test of willingness to buy a good and reveals the consumer-voter's demand for public goods." The movement implies, it should be noted, both enjoying the public goods available and paying the concomitant taxes. In his account for the activities of the local authorities Tiebout relies on the assumption that each community has an optimum size at which the average cost of the public goods is minimized. Then he says (Tiebout, 1956, p 419) "that communities below the optimum size seek to attract new residents to lower the average costs. Those above optimum size do just the opposite. Those at an optimum try to keep their populations constant."

Certainly this whole idea about competition in policy has some appeal, but there are also reasons for questions. In the rest of this chapter I will evaluate various aspects of the idea.

Deficiencies in the Market Analogy

A first point to be made is that the market analogy is halting. The clearest analogue, I think, is on the demand or customer side. Thus, in a lot of cases, it seems quite clear that people migrate from bad polities, because they are bad, and to good polities, because they are good. On the supply side, however, the analogy is much less clear. It is difficult to find correspondents to entrepreneurs and profits in the political context. This may be illustrated by a few examples.

Let us consider once again the following passage from Buchanan quoted above: "If there is an exit option, if there is a chance to leave, this necessarily imposes discipline on those who would exploit you through a political structure or a bureaucratic structure." At first we may note that the passage testifies to the tendency of citizens to move, if they can, from bad to good polities. So far everything is all right. Here, though, the passage is interesting mainly because it contains an uncommonly clear description of the incentives on the supply side. It says that the policy supplier exploits the citizens, and thus gets a profit. Now, if the citizens leave, there are two options available to the supplier. One, which is indicated in the quotation, is to make the exploitation less severe. The other is to strengthen the border control and thereby impede the emigration. The problem with the first strategy, as a market analogue, is that it does not increase the profit, rather the opposite. The problem with the second strategy is that it does not make the citizens' lives better, rather the opposite. With the incentives assumed in the example there is thus no correlation between a high profit for the policy supplier and good lives for the citizens.

Consequently the kind of profit-motive illustrated by the example does not work in a theory of policy competition. In the ideal economic market, we remember, the entrepreneurs gain their profits by developing and delivering products that the customers find attractive and therefore buy. Similarly in policy competition the profits, whatever their exact nature, should be a result of innovative policy-making that attracts citizens. The exploitation in the quoted example is rather the opposite. There the profit, to the extent that it exists, is a result of closed borders and trapped citizens. An equivalent comment may be made about Niskanen's example, quoted above. What, exactly, is the relationship between the interests of the "purely autocratic government" in that example, and the well-being of the citizens?

In Tiebout's theory there is a somewhat different problem on the supply side. In that theory, we remember, it was assumed that each community has an optimum size at which the average cost of the public goods is minimized. But do the actions of the authorities really lead in the direction of that optimal size? The problem is that in an ordinary commune, in the real world, the average costs are likely to hit the residents directly. Suppose, for instance, that we start with a population that is smaller than the optimum. If so, the residents will probably have to pay more than the optimal costs, and that is likely to induce some of the actual residents to leave the commune rather than to attract new ones, which, in turn, will cause more residents to leave, and so forth. In order to resist and reverse this mechanism, it seems necessary to leave the commune to a profit-making entrepreneur who, confident in his own judgement, sets the prices at the optimum level and suffers the losses in the short run. Certainly entrepreneurs like that are imaginable, but they do not usually appear in real communes. In fact, competition, although not far away, may be less important in Tiebout's theory than often imagined. Another indication of this is that Tiebout primarily talks about a mechanism for separating the citizens into groups according to their true preferences of various public goods. The local governments just produce different mixes of public goods for different people. This is hardly competition.

Another problem with the market analogy is that it is unclear what the competition is about. The most common idea, I think, is that the competition is about citizens. The competition may, however, also be about capital or about tax-bases in general. These latter possibilities are, I think, in general more reasonable. It is, on the whole, easier to imagine competition about capital, and about tax-bases, than about citizens. Even so, however, I believe that the incentives related to migration are usually very complex and dependent on the conditions in the individual case. This may be illustrated by a problem in Sweden caused by the large cohorts born in the nineteen-forties. These people will soon become pensioners, which means that the burden of supporting them will fall heavily on younger, and smaller, cohorts. A solution proposed is to import labor. This import is likely to increase the GNP considerably since the 40-cohorts leave behind them a lot of real capital that may be highly productive but only, obviously, if utilized by some people. This GNP increase is crucial since it may contain the resources needed for the pensioners. In order to get those resources there the immigrants must however be taxed. Even with the very high Swedish tax rates they may accept this if the wage level in their home countries is low enough. If so, they may still improve their lives a lot by moving to Sweden.

Here we thus have a case in which a polity, in a sense, tries to attract people. It is, however, more to the point to say that it tries to attract something which, when it gets into the country, becomes a fertile tax-base. The preconditions for this, in turn, is that there is an empty place in the population pyramid in the importing country, and that the original wages of those immigrating are low enough. These conditions are obviously special, complex and not omnipresent. Even if the example thus shows that there may sometimes be reasons for a polity to attract people, it obviously does not lend support to the idea of a general competition of that kind.

Still another problem with the market analogy is that the competing entities may not be as small, and thus as numerous, as sometimes suggested. Remember, for instance, that Buchanan and Tullock, in the Calculus of Consent, argued for the introduction of "market like alternatives into the political process" after having made the case for the smallest possible units of collective activity. But these small units, as we have seen, may not be as small, or as independent, as indicated by Buchanan and Tullock. They may rather be elements at the lower levels of governing hierarchies. If so, we again have a case of mistaking hierarchies for federations. Furthermore, this particular argument in favor of competition loses some of its edge.

The market analogy is thus deficient in some respects, and in particular the lack of an analogue to the profit-making entrepreneurs is a memento. Still this does not mean that the idea of policy competition is necessarily wrong. The conclusion, rather, is that the idea should be scrutinized on its own terms, and I will try to do so.

The Conditions of Variety and Mobility

Policy competition, according to it proponents, requires that two basic conditions are fulfilled - conditions which are practically always mentioned in discussions about these matters, on or between the lines. They are, for instance, obviously present in the examples given by Buchanan, Congleton, Niskanen and Tiebout in the first section of this chapter.

When the two conditions of variety and mobility are fulfilled the individual citizen can freely move between the polities and settle, perhaps temporarily, in the one he or she finds best. The idea is that these movements will improve the policy-making, but that is not necessarily so. One important problem is that people move for a lot of other reasons than political ones, and perhaps even mostly for other reasons. People may, for instance, move because they have got a new job, because they are marrying, because they can afford a more expensive living, or because they want to change something else of a non-political nature. These kinds of non-political movements are also primarily taking place within sovereign nations, and for the same reasons as political movements, namely that it is easier to cross intranational borders than international ones.

Now, if people's movements within a country are dictated more by non-political reasons than by political ones, the incentives created for local governments may be just the opposite to the ones imagined by the proponents of competition in policy. These governments, since they are the representatives of people who come and go, i e short-term residents, may rather find it rational not to bother too much about their communes. In contrast to Niskanen's claim, there may thus be a real risk that local public affairs are badly run rather than the opposite. And if local governments fail in that way, this also indicates the importance, locally, of the central government and of private enterprises.

This argument is in fact honored by both Tiebout and by Buchanan and Congleton. Tiebout thus notes that "there has been very little empirical study of the motivations of people in choosing a community" and he acknowledges that people may move for a lot of other reasons than political ones, and that this is a problem. In order to make his theory work he has to eliminate these non-political incentives. He excludes, for instance, movements dictated by the labor market by explicitly making the rather extreme assumption that "all persons are living on dividend income". According to Buchanan and Congleton (Buchanan & Congleton, 1998, p 140), "Tiebout-type mobility is often characterized as voting with one's feet. Here, location reveals that one prefers the services, amenities, and job prospects associated with one location over all others - net of moving costs. The smaller are these moving costs, or threshold costs generally, and the smaller are the variations in other nongovernmental services, the more location itself can be interpreted to imply consent." The italics, which are mine, confirm my point.


Another idea sometimes expressed by the proponents of political competition is that a federation may be very suitable for conducting policy experiments. Oates, for instance, writes that (Oates, 1999, p 1132) "(i)n a setting of imperfect competition with learning-by-doing, there are potential gains from experimentation with a variety of policies for addressing social and economic problems. And a federal system may offer some real opportunities for encouraging such experimentation and thereby promoting 'technical progress' in public policy." Besides underlining this possibility Oates also claims that "laboratory federalism" actually has proved its value in a number of cases. "There are", he writes, "a number of important and intriguing examples of policies whose advent was at the state or local level and that later became fixtures of federal policy", for instance unemployment insurance and tradeable emission rights in environmental policy.

These ideas are obviously interesting. Still there is a comment that is important, although it is basically terminological. Again it is the confusion between federations and hierarchies which is at issue. The point is that if we are really talking about conscious experimenting, the experiments have to be designed and enacted by someone, and that someone may reasonably be the national government. But if that is the case we are, strictly speaking, not dealing with a federation, since in a real federation the states are free to do what they want within their legally determined limits - they could not be given orders by the national government. Rather we are dealing with a hierarchy that, in fact, is an excellent instrument for experimentation. A case in point is the current Swedish experiment concerning the rules for selling liquor. The general rule has been that liquor stores should be closed on Saturdays. Now, however, the stores are open on Saturdays in some local communes on instruction from the national authorities. This, by intention, is a clear-cut experiment. The basic idea is to compare the effects in the two groups of local communes, the ones with closed stores and the ones with open stores.

A federation is not a proper place for experimentation in this sense. In a federal context we will rather find another mechanism for policy innovation which, although it is not experimenting in the strict sense, still is quite interesting. The characteristic trait of a federation is that the states, individually and on their own initiatives, are free to implement various policies. In such a context, other states, and perhaps the federation itself, may adopt a policy that has proved good in some state, and avoid a policy that has proved bad somewhere. This, of course, is interesting and valuable, but it is not experimenting in a strict sense. Rather we are dealing with a multi-actor, trial-error process. From the central or federal point of view, the process can be considered stochastic. The federal authorities just have to sit waiting for what might happen, they cannot steer the process. Still the mechanisms described may give rise to a lot of innovative thinking - they may be quite fertile. If so, the proponents of competition in policy would say this is likely to be a result of the competition between the states of the federation.

Present India offers an interesting example. There, the three southern states of Karnataka, Andhra Pradesh and Tamil Nadu exhibit economic growth nearly twice the national rate. The cities of Bangalore, Hyderabad and Madras in these states harbor India's booming software industry, which enjoys the benefits of free-market reforms enacted by the state governments. The enterprises continually enter into new alliances with Western companies like Microsoft, and the stock markets flourish. As a consequence, the three states are quickly becoming models for other states, and for the nation as a whole.9 State-level initiatives and referendums in the US give other interesting examples. There, successful policies pioneered by one state are often taken up in other states as well. (Moberg, 1999)

Innovation processes of the kind discussed here, in which successful policies are reproduced elsewhere and unsuccessful policies are avoided, are however also present in other contexts than in federal ones. The same kind of mechanisms may also work between independent nations. Such nations may observe each other in search of policy guidance, and they certainly do so to a large extent.

Thus, the conclusions in this section are that real experiments require hierarchies rather than federations, and that innovative trial-error processes may occur between independent nations as well as between the states of a federation.

The Framework for the Competition

I have already mentioned the idea that a federation is a particularly suitable instrument for competition, and that therefore the competition between the states in a federation may be more effective, and bring more benefits, than the competition between independent nations. One might for example imagine that rich variety and high mobility in a federation works in that direction. A memento, however, is that real federations, as we saw in chapter 3, never have been created for the purpose of competition. The competitive federation is thus, at least to a large extent, just an intellectual idea. In this section I will try to bring the comparison between competing states within a federation, and competing independent nations, a bit further.

At first a point of terminology should be clarified. Hitherto, I have spoken quite generally about competition in policy, but obviously policy may include very different matters. I will therefore make a distinction between decisions about law, i e legislation, and decisions about other things. The distinction is important since legislation is about general rules, which people should follow whenever they apply, whereas decisions concerning other matters are usually about individual things, such as the building of a school. The consequences of legislation are thus different from, and usually much more far-reaching, than the consequences of other decisions. Decisions about laws specify the borderlines between actions permitted and prohibited. Decisions regarding other things are usually about the selection of a particular action within a permitted area. From now on I shall not talk about competition in decisions about individual matters of various kinds. I will confine myself to competition in legislation, or in law. Rules about taxes, I should make clear, are considered to be laws.

So let us consider competition in law in a real federation, that is in a two-level system. A first point is that such a system has a propensity for redistributions, between the competing states, at the federal level. Such redistributions, combined with competition, may easily result in races to the bottom as described in the section "Problems with two-level systems" in the preceding chapter. The proponents of policy competition are certainly aware of this danger, and they may also prescribe cures for it. Buchanan and Congleton (Buchanan & Congleton 1998, p138) thus say that the federal government itself should be "constrained by the generality principle", which means that it is obliged to treat all constituent states equally. Even so, however, it seems obvious that the risk of this kind of redistributions is always present in a federation, whereas it is basically non-existent between independent unitary states.

Furthermore, it seems most likely that a two-level system is more difficult to understand than a uni-level system, and hence less transparent. This, of course, is bad. A legal system should be as transparent as possible. With low transparency follows system-manipulation, rent-seeking and tricky behavior in general. Migration, for instance, may very well be selective. If they are able to do so, people may like to be citizens in one state where the public social services are generous, to live in another one, for example where the climate is good, and to pay taxes in still another one where the tax rules are favorable. Such behavior is likely to thrive in a system characterized by legal variety, easy movements and low transparency. In addition to being transparent a legal system should also be as stable as possible. And again a competitive two-level system may easily lead in the opposite direction. In such an environment laws may vary more over time than in other systems. If so, this conflicts with the norm which is often expressed in Sweden, and possibly in other countries as well, that rules should be firm.

A somewhat similar argument rather focuses on mobility. A precondition for the competition in law to work, it seems to me, is that good laws attract people and that bad laws repel them. This however presupposes, at the very least, that people's decisions about movements are taken behind a "veil of ignorance". If that is not case, and if short-term interests thus are able to influence the decisions fully, the competition is certainly not only beneficial. Under such conditions a criminal is, for instance, not attracted by the good law that punishes him, but rather by the bad law that lets him go. Or, in other words, if there is no "veil of ignorance", good laws may very well repel people, and bad laws may very well attract them. But how, and that is the problem, is decision-making behind a veil of ignorance to be reconciled with easy and high mobility?

More generally, it may be said that a society characterized by the "rule of law" has rules which are transparent, firm, easy to accept behind a veil of ignorance, and unavoidable. Federal competition in law, I think, tends to lead to results being in conflict with this ideal.

Another problem with two-level systems is related to the collection of taxes. This problem may be illustrated by starting from a discussion among economists about mobile tax-bases and tax evasion. In this discussion, which was started by Musgrave (Musgrave, 1959), two distinctions are made, the one between different tax-bases according to their mobility and the other between benefit and non-benefit taxes. Starting with the latter distinction benefit taxes are taxes which can be regarded as payments for goods received or utilized, for example taxes used for providing public goods. The taxes used for macro-economic policy, e g for overbalancing the public budget, and for income redistribution, are, on the contrary, called non-benefit taxes, since they cannot be considered to be payments for anything that the taxpayer receives or uses. Now, it is argued that the incentives to tax evasion are essentially stronger for non-benefit taxes than for benefit taxes. The argument, as it seems, is that without benefit taxes there will not be any benefits either, for example no public goods. Therefore benefit taxes could use mobile tax-bases without much harm. Non-benefit taxes on the contrary, and since mobile tax-bases easily escape taxation, should only be taken from relatively immobile tax-bases. Or, in Musgrave's words (Buchanan & Musgrave, 1999 a, p 170): "Immobile land must accept whatever burden is imposed upon it. Labor, still relatively immobile, comes next while capital (and particularly financial capital) is free to move. As a result, capital can be taxed on a benefit basis only."

It should be added that the mobility of tax-bases in this kind of argument not only depends on the nature of the base, but also on the area considered. The assumption is usually (Oates, 1999, p 1125) that "... there exists a nation state with a central government, where there is little or no mobility across national borders; at decentralized levels, in contrast, economic agents, goods, and resources have significant mobility across jurisdictional boundaries with the extent of this mobility increasing at successively lower levels of government." And again the conclusion is "that decentralized levels of government should avoid nonbenefit taxes on mobile units."

This reasoning may seem plausible but it contains, I submit, a flaw in the part dealing with benefit taxes. It is, of course, true that there will not be any benefits, such as public goods, without taxes, but only at an aggregated level. From the point of view of the single individual there may still be a tempting possibility of avoiding the taxes and still have access to the public goods. This, in fact, is what the whole discussion on public goods and free-riding (Olson, 1965) is about, and this discussion thus seems ignored in the reasoning presented here. We must therefore conclude that even benefit taxes should avoid mobile tax-bases, and that they should preferably be collected centrally.

Having drawn that conclusion I have also come to my own main point here. If benefit taxes for local public goods, say public goods in the states of a federation, are federally collected, they have to be sent subsequently to the states, to whom they belong. Already for this reason the federal level thus has a certain amount of control over the financial resources of the states. This being so, it is easy to imagine that the federal authorities, when returning the money to the states, adjust the sums according to some kind of redistributive policy. That, however, brings us very close to the redistributive races to the bottom described above.

Consequently, there are some considerable problems associated with policy competition within federations. At the same time it should be noted that mobility between independent nations is not necessarily that low. Measures, which increase the mobility of human beings and of other things as well between independent nations, could in fact, to a large extent, be undertaken by such nations in ordinary non-federal cooperation, and that also happens. Rules about passports, visas, and so forth, tend for instance to be less and less severe in many countries. From this point of view, policy competition between independent nations is therefore perfectly possible.

It is, by the way, interesting to consider Buchanan's argument concerning competition in money, which was presented above, in this context. Buchanan strongly favored, we remember, a system with competitive European currencies to the Euro-system, which eventually was established. It is arguable, I think, that the Euro-system is a real two-level system in the sense that the Euro is hardly possible without a real federal level, whereas the system with competing currencies is quite realistic in a context with collaborating independent nations. "All that would have been needed", as Buchanan said, "would have been something in the treaties that would have allowed any citizen of any country in Europe to make contracts, including paying taxes, in any money that is issued by any bank in the structure. In that context, the separate central banks could still operate, and you could still have the local currencies, ...". This, as it seems, is perfectly possible in an ordinary one-level international environment, whereas the solution which finally emerged, to Buchanan's regret, is a typical two-level solution.

The conclusion of this section is thus that at least some of the beneficial effects, which often are attributed to competition within federations, may be obtained by competition between independent nations as well. On the other hand, competition within federations is associated with some severe problems, which do not appear in the competition between nations.

Competition in Law through Fission and Fusion

It should be clear by now that I am doubtful about the merits of competitive federalism. Even so, however, it may matter how such federalism is brought about. From the point of view of a single unitary nation, say Sweden, there are two totally different ways available. The one is fission, which means splitting one's own nation into a federation. The other is fusion, which means creating a federation together with some other independent nations. In the first case the mobility is there from the beginning, and the function of the fission is to create variety. In the case of fusion it is the other way round. The variety is there from the beginning and the function of the fusion is to pave the way for mobility. In this section I will compare these two possibilities.

First, fission entails legislation in many areas, which are much smaller than the homogenous cultural-linguistic areas described in chapter 2. This necessarily entails large costs for learning the laws. In a federation created through fusion there is no correspondence to this. On the other hand the external costs, or the decision-making costs, may be very high in that case. That can, however, be avoided by limiting the federal legislation to areas where there are no cultural conflicts. All arguments for these statements are already given in chapter 2. Considering the costs related to legislation we might therefore conclude that fusion is preferable to fission.

Second, the risk of central government redistributions as described in the section "Problems with two-level systems" in chapter 3, and in the section "The framework for the competition" in this chapter, is reasonably much greater in a federation created through fission than in one created through fusion. The reason is that the original level is likely to remain powerful, and faithful to its traditions, for a long time after the creation of a federation. In a federation created through fission the federal level will therefore remain strong for long, whereas the same is true for the state level in a federation resulting from fusion. In a fissioned Sweden it is therefore likely that the central government, still quite strong and faithful to its habits, would continue to redistribute resources between the constituent parts. Thus, and again, fusion is preferable to fission.

Whatever the merits of competitive federalism, I therefore conclude that a federation created by fission is hardly an attractive alternative, particularly not for a small country like Sweden.

The Creation of Good Societies

I said above that the main deficiency in the market analogy is the lack of correspondents to entrepreneurs and profits in the political context. I will now turn to the implications of this deficiency. Competition in policy relies, as we have seen, on the idea that citizens move from bad to good polities. This is often called to "vote with one's feet" and certainly people may do so. Still there is the problem with the appearance of good societies. How do they come about, and through which mechanisms? It is here that the lack of profit-making entrepreneurs is a problem. If there are no actors like that, how are the good societies, to which the individuals are supposed to move, created?

The discussion of that question can depart from some ideas about voting. The expression to "vote with one's feet" indicates that political migration has similarities with the ordinary voting at the ballot-box. The expression is, however, also misleading since it overemphasizes the similarities, and disregards the differences, between the two kinds of voting. In Hirschman's (Hirschman, 1970) terminology voting with the feet is an exit-reaction, whereas voting at the ballot-box is a voice-reaction. Hirschman maintains that if exit is easy, the voice-option is not very important, whereas if exit is difficult or impossible, the voice-option becomes important. He does however not maintain the opposite, that exit becomes unimportant if the exercise of the voice-option is easy, and for a very good reason. The two kinds of acting are very different in the sense that voting with the feet is essentially an individual act, where as voting at the ballot-box is (part of) collective action. The effects of voting with the feet are certain, total and immediate for the individual concerned, whereas the effects of ballot-box voting are, at best, uncertain, indirect, and most likely partial. From the individual voter's point of view everything depends on the extent to which other people vote in the same way as he himself. For this reason voting at the ballot-box can never substitute voting with the feet, whereas voting with the feet certainly can substitute voting at the ballot-box, at least for the individual and in the short run.

This last reservation is, however, crucial. A basic democratic idea, or perhaps even the basic democratic idea, is that a society should be built by its members. This means, in other words, that society-building is by necessity a collective undertaking. It is to a large extent dependent on stationary citizens exercising their citizenship. Therefore, and in spite of its attractions to the individual, voting with the feet does not build societies. Society-building is fundamentally dependent on the voting at the ballot-box and other essential democratic processes. The ideas of policy competition are therefore, to a considerable extent, at odds with basic democratic thinking.


My main conclusions from the discussion above about federalism and competition in government are as follows:



1There exists a fairly well developed theory about hierarchies. Main contributors to the theory of hierarehies are Simon (Simon, 1961), Tullock (Tullock, 1965), Downs (Downs, 1967), and Niskanen (Niskanen, 1971). Moberg (Moberg, 1994), chapter 7, contains a presentation in Swedish of these theories. It may also be noted that hierarchies are usually contrasted to markets rather than to federal systems. An example is Williamson (Williamson, 1975).

2It is interesting to note how well an important idea of Friedrich Hayek fits into this rationale for hierarchies. What I am thinking about is Hayek's distinction between scientific or general knowledge, and "knowledge of the particular circumstances of time and place". In making this distinction he argues that the importance of the latter kind of knowledge is often seriously underestimated. He writes (Hayek, 1945 and 1980, p 80) that "practically every individual has some advantage over all others because he possesses unique information of which beneficial use might be made, but of which use can be made only if the decisions depending on it are left to him or are made with his active cooperation."

3There are many works dealing with federalism and its various aspects. A main reference in this essay is Riker (Riker, 1964). Another important reference is Oates, 1999. Federalism as a mechanism for providing institutional competition is discussed by Berggren (Berggren, 1997), which also contains further references.

4A confederation is a group of sovereign states with a common purpose, usually defensive or economic, allied by a treaty. In a confederation there is thus no federal level with a federal government (Bealey, 1999, the entries "confederations" and "federal government").

5Riker, in fact, again and again emphasizes that there is no connection between federalism and freedom.

6For a further explanation of the concept, see Buchanan & Tullock, 1962.

7See for instance Gerholm, 1999, or, in Swedish, Gerholm, 1998.

8In spite of its name the Swiss confederation is, according to the established terminology, a federation and not a confederation (Bealey, 1999, the entry "confederations").

9The example is taken from the article "Reform grabs more seats in Delhi - Tech-savvy free marketeers won big in the October election" in Business Week, October 25, 1999.


Bealey, F, 1999, The Blackwell Dictionary of Political Science. Oxford: Blackwell Publishers Ltd, 1999.

Berggren, N, 1997, "Social Order through Constitutional Choice: A Contractarian Proposal" in Berggren, N, Essays in Constitutional Economics, Stockholm School of Economics, 1997.

Buchanan, J M & Tullock, G, 1962, The calculus of consent, logical foundations of constitutional democracy. Ann Arbor: The University of Michigan press, 1962.

Buchanan, J M & Congleton, R D, 1998, Politics by Principle, not Interest - Towards Nondiscriminatory Democracy. Cambridge: Cambridge University Press, 1998.

Buchanan, J M & Musgrave, R A, 1999a, Public Finance and Public Choice. Cambridge, Massachusetts: The MIT Press, 1999.

Buchanan, J M, 1999b "Discussion of political science and economics: three research programs in constitutional political economy", in Competition & Cooperation - Conversations with Nobelists about Economics and Political Science, (ed J E Alt, M Levi & E Ostrom). New York: Russell Sage Foundation, 1999.

Downs, A, 1967, Inside Bureaucracy. Harper Collins Publishers, 1967.

Friedman, M & R D, 1998, Two Lucky People - Memoirs. Chicago & London: Chicago University Press, 1998.

T R Gerholm (ed), 1998, Klimatpolitik efter Kyotomötet. Stockholm: SNS förlag, 1998.

T R Gerholm (ed), 1999, Climate Policy after Kyoto. Bentwood: Multi-Science Publishing Co Ltd, 1999.

Hayek, F A, 1945, "The Use of Knowledge in Society", in the American Economic Review, XXXV, September 1945. Also reprinted in Hayek, F A, Individualism and Economic Order, the University of Chicago Press, 1980.

Hirschman, A O, 1970, Exit, voice and loyalty: responses to decline in firms, organizations, and states. Cambridge (MA): Harvard University Press, 1970.

Lijphart, A, 1999, Patterns of Democracy, Government Forms and Performance in Thirty-Six Countries. New Haven: Yale University Press, 1999.

Moberg, E, 1994, Offentliga beslut - en introduktion till den ekonomiska teoribildningen inom statsvetenskapen. Lövestad: Mobergs Publikationer AB, 1994.

Moberg, E, 1999, "Positive and normative aspects of direct democracy - the case of Sweden in a general context", in N Berggren, N Karlson & J Nergelius (ed), Makt utan motvikt - Om demokrati och konstitutionalism. Stockholm: City University Press, 1999.

Musgrave, R M, 1959, The Theory of Public Finance. New York: McGraw-Hill.

Niskanen, W A, 1971, Bureaucracy and Representative Government. Chicago, Aldine Publishing Company, 1971.

Olson, M, 1965, The Logic of Collective Action, Public Goods and the Theory of Groups. Cambridge, Harvard University Press, 1965.

Oates, W E, 1999, "An Essay on Fiscal Federalism", in Journal of Economic Literature, XXXVII, September 1999.

Riker, W H, 1964, Federalism - Origin, Operation, Significance. Boston and Toronto, Little, Brown & Company, 1964.

Simon, H A, 1961, Administrative Behavior, A study of Decision-Making Processes in Administrative Orgnization. New York, The Macmillan Company, 1961.

Tiebout, C M, 1956, "A pure theory of local expenditure", in Journal of Political Economy, 64, October 1956.

Tullock, G, 1965, The Politics of Bureaucracy. New York, University Press of America, 1987.

Williamson, O E, 1975, Markets and Hierarchies: Analysis and Antitrust Implications. New York, The Free Press, 1975.