Those interested in this paper may perhaps also be interested in the book "Towards a Science of States: their Evolution and Properties"


Erik Moberg:

Positive and normative aspects of direct democracy - the case of Sweden in a general context

ã Erik Moberg and the City University of Stockholm


1. Introduction
2. Referendums in non-constitutional contexts
3. The constitutional ideals of directness and unanimity
4. The relevance of the nature of the issue
5. The formation of proposals
6. The possibility of non-voting
7. Political parties and representation
8. The constitution and the political parties
9. US experiences
10. Swiss experiences
11. Italian experiences
12. The case of Sweden - experiences
13. The case of Sweden - proposals for change

1. Introduction

In a pure, direct democracy all public decisions are, in principle, taken collectively by an assembly consisting of all the adult citizens in the society. In a representative democracy, on the contrary, the citizens elect, in one way or another, a smaller number of representatives who, in their turn, make the public decisions. All modern democracies are, as we know, mainly representative, or indirect, even if elements of direct democracy are also used now and then. The most important of those elements, and the ones to be discussed in this essay, are the referendum, and the initiative.1

Table 1 gives a rough picture of the use of referendums in the world (including a number of initiatives in Switzerland). Its shortcomings are essentially that it contains no recordings after the mid 1990s, although quite a number of referendums have occurred since then, and that it is confined to national referendums. It should therefore be added that initiatives as well as referendums are frequent, and important, not only in the Swiss cantons, but also in US states, in particular California, Oregon, North Dakota, and Washington, and in the Australian states. Five major democracies, namely India, Israel, Japan, the Netherlands, and the United States have never had a nation-wide referendum.2 The use of direct democracy is thus a fairly widespread, but also rarely used, phenomenon. Furthermore the use varies considerably from country to country. Switzerland and the US states use direct democracy much more than any other polities.

Are there, we may ask, any plausible explanations for this order? Are there, for example, any factors which explain why some countries use the institutes of direct democracy so extensively, while other countries abstain from them almost altogether, or even completely? In addition to these positive questions there are also important normative ones. Rather than asking for explanations we may ask for rationales for various orders. Are there any reasons which, on the whole, make representative democracy preferable but which, nevertheless, in certain specific cases, call for elements of direct democracy? If so, which are these reasons?

Table 1: Number of national referendums since 1945 to the mid-1990s.3

In this essay I intend to discuss these questions, the positive as well as the normative ones, and the ultimate purpose is to arrive at conclusions about Sweden in particular. Sweden has, as we know, a referendum institute, and, over the years five national referendums have been taken. Is this institute, we may ask, working well, or should it be changed in some way? Furthermore, it may be asked if there is a place for a national initiative institute, which currently does not exist, in the Swedish political system? For the discussion of these questions it is, I believe, expedient to look at Sweden in a broad perspective, taking into account world wide experiences of direct democracy. These experiences, as such, are however hardly enough. It is also, I think, essential to consider some relevant theoretical topics. Thus it is important to discuss some general qualities of direct and representative democracy respectively. Furthermore, it is necessary to say something about the character of political parties in different types of representative systems, since these qualities are highly relevant for the workings of the systems, and for the systems' compatibility with elements of direct democracy. I shall accordingly treat these topics before turning to the historical experiences in a few particularly interesting countries, and before turning to Sweden in the final sections. Hence the title of the essay.

2. Referendums in non-constitutional contexts

In this essay, because of our interest in Sweden, we are mainly concerned with situations in which the elements of direct democracy are used according to established constitutional rules, or at least in ways which are compatible with such rules. In such situations direct democracy may be said to be used in a constitutional context. It is, however, important to realize that one of the elements, the referendum, has a long and important history in contexts where there is no commonly accepted, or democratic, constitution at all, or, in other words, in non-constitutional contexts. We may therefore deal briefly with this topic before starting the main discussion about the constitutional use of referendums and initiatives.

In societies not characterized by constitutionalism and rule of law it is often important for leaders, or prospective leaders, to consolidate their positions by various proofs of legitimacy. Ad hoc referendums may be useful instruments to serve that purpose. A possible stratagem is to present the referendum as a vote of confidence and thereby, besides enforcing an affirmative answer to the proposition put on ballot, achieve the requested legitimizing support. A modern example is the proposition which the Chilean people were asked by Pinochet to support in 1978: "In the face of international aggression unleashed against the government of the fatherland, I support President Pinochet in his defense of the dignity of Chile, and I reaffirm the legitimate right of the republic to conduct the process of institutionalization in a manner befitting its sovereignty."4 Among Pinochet's colleagues using this and similar techniques are Napoleon I, Napoleon III, Mussolini, and Hitler. The examples set by the French rulers explain the term Bonapartism, meaning "plebiscitary democracy as the servant of authoritarian regimes".5

This historical record has made many people hold that the referendum institute, rather than being an instrument of democracy, is primarily a dangerous weapon in the hands of dictators seeking popular support for ruthless policies. When, in 1945, for example, Winston Churchill proposed a referendum in England about the prolongation of the wartime coalition, the labor leader Clement Attlee retorted that: "I could not consent to the introduction into our national life of a device so alien to all our traditions as the referendum, which has only too often been the instrument of Nazism and Fascism. Hitler's practices in the field of referenda and plebiscites can hardly have endeared these expedients to the British heart."6

But even if dictatorships provide important examples of non-constitutional contexts, they are obviously not the only ones. To the non-constitutional category we may also refer polities in various kinds of transitional phases. A new societal order may have to be built since the old regime has collapsed, or is considered ripe for collapse. In such non-constitutional situations referendums, although not stipulated in any constitution, may be crucial for giving legitimacy to the change. They can, for example, be instrumental in enhancing the passing of the old order, or in establishing the new one, which may very well be a democratic one. The countries which were previously parts of the Soviet empire have seen many such referendums in later years - some of them recorded in table 1.7 Perhaps we may also refer the referendums in France under de Gaulle, most of those in table 1, to this category. The referendums we are dealing with here are, as I mentioned above, often used as votes of confidence. It might therefore be added that de Gaulle in 1969, and Pinochet in 1988, resigned after defeats in referendums.

The whole topic of referendums in non-constitutional contexts is thus a complex one. Certainly it is much too simple to associate referendums in general with totalitarianism, but it is also, as we have now seen, too simple to make that association for referendums in non-constitutional contexts in particular. True, referendums may be used by dictators for enhancing their power, but referendums may also be of crucial importance in situations where new democracies are emerging. Even in non-constitutional contexts they may thus serve democracy as well as totalitarianism. With that conclusion, and considering our interest in Sweden, we have to leave the non-constitutional contexts. Henceforth I will deal with cases where the elements of direct democracy are used in accordance with established constitutional rules.

3. The constitutional ideals of directness and unanimity

Going back to basics, and starting with simple examples or even models, we may consider a direct democracy using the unanimity rule. Let us furthermore assume that the citizens always vote for proposals which favor them, and against proposals which harm them. We can call this an assumption of spontaneous voting - in other words it means that the voting is forthright and void of tactical aspects. Under these conditions, it is easy to see, proposals which favor all citizens, according to their own subjective preferences, and only such proposals, will pass. In welfare terms this means that all Pareto-improving proposals which are presented, and only such proposals, will pass.8 All proposals which are not Pareto-improving will be rejected. These qualities of a direct democracy using the unanimity rule are obviously attractive, but, alas, hardly achievable in reality, at least not completely. Nevertheless this ideal type of democracy, even if not achievable, is a good starting point for our discussion. Exactly why, we should ask, are its qualities evading realization, and what can be done to get as close as possible to the ideal?9

A first problem is that decision-making involving a whole population including millions of individuals is necessarily extremely cumbersome. It takes an enormous amount of effort to activate all the people, to make them consider various proposals, to make them adjust their ideas to each other if necessary, and so on. In the modern, technical jargon it is said that the transaction costs associated with the decision making are much lower when the group is small than when it is big. This fact thus speaks strongly in favor of representation - it is better to let a small group of representatives make the decisions rather than to engage the whole population. According to this argument the important difference between a group of representatives and the population at large is thus the size - no qualitative differences of any kind are presumed. The decision-making group can therefore be considered as a mini-people which is representative, in a statistical sense, of the real people. Hence the representatives could, in principle, be selected by sampling.

There is, however, also another argument in favor of representation which rather emphasizes the qualitative differences between the decision-makers and the citizens in general. According to this argument the people at large benefits from having a small élite making the public decisions for, and on behalf of, the whole population. The members of that élite may be considered to have acquired their skills by training, or by birth - in a society characterized by various sorts of specialization they have become specialists in public decision-making. In a community where these ideas prevail the selection of representatives by sampling is, of course, excluded. Rather the citizens have to identify, and choose, representatives which they consider to be qualified to represent them. That task is accomplished in general elections.

Both arguments for representation have considerable validity.10 As for the first one it is evident that a system of representation reduces at least some transaction costs drastically. The argument, although to a large extent sound, is however not as conclusive as it may seem in the first instance. As we shall see below (the section "Political parties and representation") some transaction costs may in fact increase in a representative system, and furthermore, as we shall also see (the section "The formation of proposals"), low transaction costs are not necessarily welcome. The second idea about a representing expert élite is also basically sound. The problem is to know in which situations, and in which respects, expert knowledge can be admitted.

For the discussion of this problem we may start with the liberal, individualistic contention that every man is, himself, the best possible expert on his own feelings or values. Thus, if a man says that a particular state of things is good enough for him, that holds true, irrespective of what others might think. What makes the Pareto-criterion so patently important is precisely that it honors this fundamental contention - the welfare of the society is expressed in terms of all the citizens' own, individual evaluations of their own predicaments. But if this is so, how can there be a role for an expert élite at all?

The reason, it seems to me, is that decision-making requires more than knowledge about individual evaluations of various societal states. It also requires knowledge about how various desirable states are likely to be reached, and undesirable ones avoided. In other words, a knowledge about societal mechanisms, for example the different kinds of mechanisms which are studied in the social sciences, is relevant. Such knowledge is, of course, of expert nature and not at every man's disposal.11 It is thus an advantage of a representative system that it accommodates this kind of expertise.

So let us assume that the ideal of directness is abandoned at least to some extent, and that representatives are used, in one way or another, but that the other ideal, the unanimity rule, still reigns. Would that mean that all problems are solved? No it does not - the unanimity rule as well is impaired by some problems.

At first some people, whether citizens or representatives, are likely to shun spontaneous voting in favor of tactical voting. They may vote against, and thus block, even proposals favoring themselves or their constituencies. By such behavior they may hope to get the proposals changed to become still more favorable, and thus to extract some extra benefits. If the unanimity rule is strictly applied such obstructive behavior is of course extremely tempting, and extremely likely, since an individual can engage in it single-handedly. If, however, a qualified majority is substituted for unanimity such tactical behavior becomes more difficult, and thus more unlikely, since some kind of coalition-building will be required. From this point of view there are in fact good reasons for requiring a qualified majority for an affirmative decision, rather than unanimity.

To this we may add a second reason for requiring a qualified majority rather than unanimity, namely the wish to let some proposals which are Kaldor-Hicks-improving, although not Pareto-improving, pass. True, a Kaldor-Hicks-improving proposal, by definition, can be turned into a Pareto-improving proposal by the addition of a proper scheme of side-payments, but in practice it may be difficult for those preparing the proposals to find all side-payments which are necessary to do the trick. They may have to be satisfied with a scheme for side-payments which, even if not perfect still makes things better, and thus turns an original Kaldor-Hicks-improving proposal into a "better" Kaldor-Hicks-improving proposal (fewer citizens harmed and more benefiting). If so there are good reasons for letting the final Kaldor-Hicks-improving proposal pass, but that will require that a qualified majority rule rather than the unanimity rule is used.

Thus it is obvious that a pure direct democracy, using the unanimity rule, is not workable. It is necessary, at least to some extent, to introduce representatives, and thus to rely on indirect, rather than direct, democracy. It is also necessary to depart from the unanimity rule in favor of some kind of majority rule, or perhaps a set of such rules. Once the principles of directness, and of unanimity, are abandoned various new risks are however also introduced.

The main risk with a representation system is that it fails, that the representatives represent other things than their proper constituencies. As we shall see below (the section "Political parties and representation"), such failure may sometimes be due to qualities of the democratic system, and thus occur even if the politicians act with the best of intentions and as honestly as they can. In addition to this it is however also important to note that a politician, in addition to representing other people, always has interests of his own as well. The politician may therefore abstain from representing his proper constituency, and rather use his position to improve his own lot. He may, for example, act on behalf of various moneyed interests in exchange for favors given to himself. If so the politician's behavior is for sale, and we are dealing with outright corruption.

In welfare terms wavering representation is always likely to be destructive. In milder cases, where the politicians are just not representing their constituencies properly, without favoring themselves, it is still likely that some citizens are favored at the expense of others. If so, since incentives are likely to be impaired, and thereby growth, we are dealing with Kaldor-Hicks-deteriorations, at least in the long run. In more severe cases, where politicians are using their positions to enrich themselves, the consequences are likely to be much more serious. Here essential parts of the social fabric may be at issue, the society may begin to disintegrate, and political violence may start spreading. This, obviously, implies Kaldor-Hicks-deteriorations. When judging, or designing, a representative system it is important to be aware of these hazards.

The main risk in using a majority rule rather than the unanimity rule, is exploitation. In principle it is possible for the people belonging to the decision-making majority to tax the people in the remaining minority and share the spoils. Such exploitation is likely to be destructive in the long run. Since many citizens will feel offended, incentives will be distorted, and growth impeded, which is tantamount to Kaldor-Hicks-deteriorations. Another similar risk is the one of majority harassment of various cultural or ethnical minorities. Perhaps these risks may be diminished by a careful selection of the rules by which the representatives are elected, so as to make sure that they are endowed with a reassuring amount of impartiality.

We can thus conclude that a pure, direct democracy, using the unanimity rule, although ideal in one way, is not practicable. For at least some purposes it is necessary to introduce representatives, and it is also necessary, at least partially, to abandon the unanimity rule. These modifications, although necessary, do however also introduce new risks, in particular the risks of corrupt representatives, and of exploitation of parts of the electorate. These risks are of course also present in the democratic systems of the real world in which, whatever the reasons, representativity is always a dominating feature, and where the unanimity rule is never used.

Bogdanor, when considering the potential good of the institutes of referendum and initiative says that the referendum "is a device intended to repair the legislature's sins of commission", while "(t)he initiative, by contrast, is a device intended to repair the legislature's sins of omission."12 Using this distinction it is clear that the risks of failing representation, and of exploitation, when realized, are sins of commission. Obviously it is important to mitigate these risks and, agreeing with Bogdanor, I shall argue that this is an interesting and proper task for referendums. Before turning to that discussion I shall, however, deal with a few more topics of a general nature.

4. The relevance of the nature of the issue

So far I have discussed the questions of representation, and of rules for decision-making, without considering the nature of the issue to be decided. This matter, however, is important and the basic arguments will be differently assessed for different kinds of issues. Here I will illustrate this relationship by discussing two categories of issues which, perhaps, may be considered as extremes, namely local issues and constitutional ones.

Local issues are often of a practical and down to earth nature rather than abstract or technical. It therefore seems reasonable to assume that such issues, in contrast to national ones, on the average require comparatively little expert knowledge for their settlement. In their homes people decide about everything themselves - about the use of the rooms, about the acquisition and placement of furniture, and so on. In national affairs they usually find it expedient to delegate most decisions to elected expert representatives. Local issues may be considered as something in between. The construction and use of public buildings and streets, for example, may have more in common with home affairs than with national affairs. If so, one of the arguments in favor of representative democracy, namely the need for expert knowledge, is comparatively weak in local politics. Thus there may also be more space for direct democracy in local than in national politics.

The arguments in favor of referendums and popular decision-making in constitutional issues are of a very different nature. Until now we have only, so to speak, discussed the game itself, whereas constitutional issues are about the rules of the game. This fundamental distinction between the game and the rules of the game, or between applied decisions and constitutional decisions, is strongly emphasized by, among others, Buchanan and Tullock.13 The main conclusion which they draw from this distinction is that constitutional decisions, in principle, should be unanimous, whereas qualified majorities could be used for applied decisions.

It may also be argued that the acceptance of the constitutional rules is, in a sense, equivalent to the acceptance of citizenship which, according to individualistic and liberal principles, should be free. Different aspects of these matters have been discussed by, among others, Hirschman, Niskanen, and Tiebout.14 Hirschman, as we know, distinguishes between two fundamental types of critical reactions, namely voice and exit. He also contends that the need for voice is small if exit is free, or that "the presence of the exit option can sharply reduce the probability that the voice option will be taken up widely and effectively".15 Niskanen expresses the same thing when he writes: "Even the most autocratic government should have to be approved by every resident if the costs of moving were zero."16

The implication of this, since free exit is an ideal, is that the constitution has to be accepted by every single, individual citizen. The way to achieve this is to submit constitutional matters to referendum, requiring unanimity for affirmative decisions. In practice, however, and because some people may block a proposal they like for tactical reasons, it may be necessary to accept a high qualified majority. These ideas are sometimes also expressed in terms of the concept of legitimacy. In constitutional matters in particular, it is said, there is a demand for, or a need for, legitimacy, which is taken to be equivalent to wide-spread and well founded popular support.

The need for legitimacy is often felt by real decision makers, as demonstrated by the great number of referendums on constitutional topics. For the presentation of these matters it is useful to differentiate between three kinds of constitutional changes.

Firstly there are the very important cases when several areas (or societies), with basically separate jurisdictions, are made more dependent of each other, or when areas (or societies) which basically belong to the same jurisdiction are made more independent of each other. For short, we may talk about jurisdictional fusion, and fission, respectively. Here are some examples of such referendums:17

Secondly there are the cases where a whole new constitution, or constitutional philosophy, is to be introduced. Examples of referendums in such situations are the establishment of the present constitutions in: Thirdly, we have the cases where just parts of a constitution are to be changed. Even here referendums are often used. In Italy and New Zealand referendums have, for example, been instrumental in bringing about new electoral laws. In Italy the old electoral law of 1948, basically stipulating proportionalism, was abrogated by a referendum in 1993, and thus the representative system was forced to bring about a new law, based on a mixture of proportionalism and plurality?23 The change in New Zealand went, so to speak, in the opposite direction. In 1993 a majority of 54 per cent favored a change from a first-past-the-post-system to proportional elections. In England, where such a change is also discussed, the present government, if it is going to propose a change, has committed itself to submitting the issue to a referendum.

Going back to the former distinction between constitutional and non-constitutional contexts it is important to note that changes of constitutions, and the use of referendums in such cases, obviously may, and often do, occur in constitutional contexts. Democratic constitutions, as we know, usually contain rules for how the constitutions can be modified, and when those rules are applied we are dealing with constitutional change in a constitutional context. But this does not preclude that some of the examples, in the first two categories above, represent constitutional changes in non-constitutional contexts. Or, since the distinction between constitutional and non-constitutional contexts is sometimes vague, they may at least be considered as borderline cases.

5. The formation of proposals

Another distinction which is important when discussing the properties of direct and representative democracy is the one between the initial formation of proposals, and the subsequent voting on the proposals. The distinction is essential since the formation of proposals, in particular, is a task which requires a lot of reflection and accuracy, and which is therefore likely to benefit not only from being done by small groups of people, but also from being done by experts. Both the quantitative, and the qualitative, arguments in favor of representation are thus particularly applicable to the proposal-forming part of the decision-making. One reason is that a proposal may have to be scrutinized in various ways, for example with respect to its compatibility with other laws, and in particular with respect to its constitutionality. Another reason is that a voting-procedure making it likely that Kaldor-Hicks-improving proposals will pass is not enough for achieving good results. It is also necessary to have a mechanism which makes it likely that the proposals presented in fact represent Kaldor-Hicks-improvements, and even that most existing potential Kaldor-Hicks-improving proposals are turned into actual proposals. To the extent that this latter requirement is not fulfilled we shall have, in Bogdanor's terminology, sins of omission. Ideas which are potentially beneficial are not raised at all.

When discussing these sins of omission it is important to note that Kaldor-Hicks-improving proposals may be of different kinds. One possibility is that they are simple proposals, benefiting large parts of the population or the entire population. A possible example may be a proposal about improved citizen's rights of some kind. In order to be likely to present such proposals, the proposing actor shall, as it seems, be endowed with a considerable amount of impartiality, and probably also with broad-mindedness and imaginative power.

There are however also Kaldor-Hicks-improving proposals which are complex rather than simple, and which therefore require, for their presentation, hard and dedicated work from the proposing actor. For the execution of such essential and difficult tasks a small group, or groups, of qualified people is required.

Consider, for example, the proposal illustrated in figure 1. The proposal is of great importance to the members of S1 whereas the members of the other subsets are only marginally disturbed by its implementation. The proposal may for example be that the members of S1 shall be allowed to educate their children, according to their own ideas, in schools which differ from those in the rest of the society. It is thus reasonable to think that the members of S1 have very strong feelings about the proposal, and that the other citizens hardly care. The value of the proposal for S1 is much greater than the losses for all other subsets together. But even so, and assuming spontaneous voting, it is obvious that the proposal, taken by itself, would be defeated in a ballot according to the majority rule. In such a ballot the intense feelings of the members of S1 do not count. The only thing that matters is that the number of members of S1 are fewer than those of S2-S7.

Figure 1

There are, however, various ways of making strength of feelings count, and this is where an active proposal-making actor has a role to play. The proposal in the figure is, according to our assumptions, Kaldor-Hicks-improving, and thus there exists, by definition, a scheme for side-payments, from S1 to S2-S7, which, together with the proposal, makes everybody better off. The proposal making actor can, for instance, work out and present such a Pareto-improving combination.

There may, however, also be other ways of handling the proposal in the figure. Assume, for example, that not only S1, but also each of the subsets S2-S7, carries some kind of interest which is of great importance for the subset itself, but is only vaguely unpleasant for the other subsets, and that these interests are presented as proposals. Now, if there were to be separate decisions for each of these proposals they would obviously all be refuted. If the seven proposals could be linked into a single proposal, that proposal would however pass. Obviously the amalgamated proposal benefits all the citizens in the society. Again, however, the linking presupposes an appropriate organization, such as a small expert group, for its fulfillment.

A small, active organization for making proposals can thus be of great value, and sometimes it is necessary. A requirement for the organization to work in that way is, however, as already mentioned, that it is reasonably impartial. If, on the contrary, the organization is partial rather than impartial, it may use its power to design proposals for other purposes. Skillful groups can thus prepare proposals, for example by linking, which favor themselves, or their constituencies, at the expense of others, and which in that sense are destructive. If so, we are dealing with sins of commission.

An instructive case is illustrated in figure 2 where we consider a society consisting of five subsets of equal size, which are proportionally represented in the legislature. Decisions are assumed to be taken by the simple majority rule. Now, assume that the representatives of the subset S1 make the proposal P1 that all people except the ones in subset S1 shall be taxed, say by 25 monetary units each, and that the money shall be transferred to the members of S1, who thus get 100 monetary units each. This proposal, voted on in isolation, will obviously only get 20 per cent of the votes in the legislature and is thus rejected.

Figure 2

Let us now continue by assuming that the representatives of the other subsets make similar proposals for their subsets. That gives us the five proposals indicated in the figure. Again, each of these proposals, voted on in isolation will obviously be rejected. But if some of the proposals are linked, and thus not voted on in isolation, things may become different. Imagine, for example, that the three proposals P1-P3 are amalgamated into a single proposal. If so that proposal will be supported by 60 per cent of the legislators in the assembly, and thus it will pass. The result is the transfers shown in the last line in the figure. The majority S1-S3 exploits the minority S4-S5.

Obviously there are transfers, for example favoring sick or poor, or otherwise suffering people, which may be considered as just and fair by most citizens, but this is not the topic at issue here. Here we are discussing outright exploitation which obviously also exists. The example illustrates the important idea that a number of minorities may sometimes unite to form a decisive constellation, which exploits the losers in some way, for example by taxing them and sharing the spoils. Distorted incentives, and thereby slower growth and Kaldor-Hicks-deteriorations, are the likely results.

Before leaving the example about exploitation one further comment is appropriate. In economics, as we know, low transaction costs are considered to be a universal good. The reason is that economic agreements, disregarding external effects, are governed by a quasi unanimity principle, and thus Pareto-improving. Under such conditions, the smaller the costs are for reaching the decisions, i e the transaction costs, the better it obviously is. In politics, however, things are different. There low transaction costs is not necessarily a blessing, and the reason is the universal use of majority rules of various kinds and thereby the omnipresent possibilities for exploitation. The linking of proposals is for example a complicated process which benefits from low transaction costs. But linking, as our example shows, may very well be destructive.

In this essay the whole topic of formation of proposals is important since referendums tend to be about single, simple, unlinked issues. Sometimes, as we shall see later on, it is even constitutionally stipulated that referendums should be about single issues. All of this may work well as a corrective against destructive linking. On the other hand it is also important to note that linking may sometimes be constructive, as illustrated by the first example above, and therefore it is important that the potential for such linking is kept alive. That purpose is perhaps best served by a representative system, including fairly impartial actors with a proposal-making capacity. There are, however, also Kaldor-Hicks-improving proposals which are simple rather than complex. In order to bring such proposals to the ballot, and thus to avoid sins of omission, an institute of initiative may also be helpful.

6. The possibility of non-voting

Hitherto, in particular when discussing spontaneous voting, I have tacitly presumed that all citizens vote. But obviously that is not necessarily the case. The voters may abstain from voting as well. For example, voters who are just insignificantly harmed by a proposal, may stay at home rather than taking the trouble of voting against it. This possibility is of great importance for the effects of initiatives and referendums. Consider, for instance, the proposal P1 in figure 2 above. Assume at first that that proposal is put on ballot as a result of a popular initiative, carried through by people favored by the proposal. Assume furthermore that the proposal passes in the subsequent referendum since those benefiting from it support it by voting, while most of the other ones don't bother at all. Such things may happen, and if they do, it is of course not desirable. We are dealing with exploitation, or Kaldor-Hicks-deteriorations.

In the following I will argue that logrolling, and decision-making, according to the total pattern in figure 2, is likely, and particularly likely in a parliamentary, proportional setting. I will also argue that referendums, in which the proposals P1-P3 are presented one by one, may be used for eliminating such outcomes. All of this, I think, is true. The outcomes may actually be eliminated. It may, however, also be the other way round. If enough citizens abstain from voting the referendums may certainly fail to reject the proposals P1-P3. Indeed, a constitutional setting which allows for both easy initiatives and referendums, may, for the reasons given, be worse than an ordinary parliamentary, proportional setting in bringing exploitation about.

One possible way of avoiding accidents like this is to introduce a quorum rule.

7. Political parties and representation

In a representative democracy there will always be political parties. The reason is that the positions as representatives are likely to be contested, and in order to succeed in the competition the candidates will find organizations of adherents, that is political parties, expedient or even necessary. But even if parties thus are always present, they are almost never mentioned in the constitutions. Still, the parties are so important for the functioning of a political system, and so relevant for the properties of its output, that a constitutional analysis which does not explicitly, and in reasonable detail, deal with the parties, is usually bound to fail.

In this study, where we are interested in direct democracy, a first important aspect of political parties is that they may make representation inadequate. In this section I will present three mechanisms impeding representation, and thereby strengthening the case for direct democracy.

The first mechanism is illustrated in figure 3. There we see three parties along a left-right scale. Obviously these parties facilitate the handling of issues of a left-right character. For such issues the transaction costs will be much lower than if there were no parties at all. For other issues the parties may however have a very different impact. Consider, for example, issues for which people's opinions tend to depend on whether they are living in cities or in the countryside, and for which the ideological left-right dimension is largely irrelevant. For such issues the three parties may very well raise the transaction costs rather than the opposite. The reason is simple. The decision-making requires that urban people, and rural people, somehow emerge as identifiable actors in order to be able to sit down and negotiate. Each of the parties P1-P3 may however find such tendencies party-splitting and may therefore suppress them. Such suppression, if it occurs, is of course an obstacle to be overcome and thus an addition to the transaction costs.

Figure 3

Another mechanism which makes political parties impede representation has been observed by Harold Demsetz. In an interesting paper he argues "... that representative democracy can yield outcomes different from those suggested by assuming that competition between political parties result in voter sovereignty."24 Demsetz reaches his conclusion by comparing political parties with private business firms. The people active within a firm are usually mainly interested in profit maximizing, and feel quite free to adjust the product in order to reach that goal. In a political party this is substantially different. The products of the party, that is its political ideas and its program, and its candidates are valuable in themselves for those working in the party, and they are therefore not prepared for all the alterations which might give more votes. The product is, in Demsetz' words, laden with "amenity potential". Therefore "a political party typically holds more stubbornly to its product mix than does a business firm". This, in combination with the typical indivisibility of political goods, paves the way for Demsetz' conclusion about the voters' limited influence, and about the representatives' deficient representativity.

A third mechanism has its ground in the complex process by which a voter in a general election decides which candidate or party to support. In the general case we may assume that the voter considers the competitors' positions on various ideological or interest-related issues, the politician's personalities, forms an opinion of the importance of these different aspects, and so forth. By weighing all these aspects together the voter finally reaches, we may assume, a decision how to vote. We may call this kind of voting compound voting. The contrast is voting on a single issue, as in a referendum. In that case the voting is straightforward rather than compound. The citizen just has to vote his or her position on the issue to be decided, and that is that.

The fact that the voters, in compound voting, weigh together a lot of different considerations in order to arrive at their decisions means, for the competitors, that they to a considerable extent must be ignorant about the nature of the support they get in an election. They will have no solid knowledge of why people voted for them. This may also be expressed by saying that the competitors, to some extent, are unclear about the mandates given to them by the voters, or ignorant about what they represent. Compound voting and unclear mandates are thus the two sides of the same coin.

This distinction between compound and straightforward voting is sometimes ignored by political scientists. Gilljam has for instance said, as an argument against referendums in Sweden, that "it is difficult to see why a referendum with a turnout of 75 per cent is a better expression of the people's will than a decision in the legislature about the same issue, taken by representatives appointed in general elections with a turnout of 90 per cent".25 One important difference between the two decisions, which Gilljam overlooks, is that the referendum explicitly is about the issue raised, and only about that issue, whereas the decision in the legislature is taken by representatives with necessarily unclear mandates. Certainly the issue was at most one of several debated in the campaign preceding the general elections, and perhaps it was not then pondered at all.

8. The constitution and the political parties

Parties, as we have seen, are always present in representative systems. Even so, however, they have, to some extent, different functions, and different properties, in different types of representative systems. Furthermore the introduction of elements of direct democracy will always affect the parties and party systerns. In the words of Bogdanor: "The general effect of the referendum is to weaken parties since it enables the voter to separate his or her opinion on issues from party allegiance."26 For a proper understanding of the impact of elements of direct democracy it is necessary to assess these influences in various representative systems, one by one.

I shall start by assuming that a country's party system to a large extent is determined by two constitutional elements, namely the rules for appointing the executive, and the electoral laws.27 There are two principal methods for appointing the executive, of which one is used in parliamentary systems, and the other in presidential systems. According to the parliamentary method the people first elects the legislature, which then, in turn, appoints the executive. Furthermore, in a purely parliamentary system the executive can remain in office only as long as it enjoys the confidence, or support, of a majority in the legislature. This requirement is therefore often referred to as the parliamentary principle. The presidential method, on the contrary, means that separate elections are held for appointing a president and thereby also the rest of the executive.

A parliamentary system depends, for its functioning, on the existence of stable, centralized and cohesive political parties in a way that a presidential system does not. The reason is that the parliamentary support, in order to be reliable and lasting, cannot be anonymous. A support expressed by an ad-hoc, transient majority of individual members of the legislature cannot, it is easy to realize, have much value. The support has to be expressed by a few stable and identifiable actors, which, in effect, means political parties.

A parliamentary system, however, is not only dependent on stable, centralised and cohesive parties; conversely it also gives strong incentives for the formation of such parties, and sometimes also for forming big parties. The reason is that those properties enhance a negotiating party's credibility and reliability and thereby its chances to become a member of the executive, a membership which is often quite attractive, or even lucrative. This attractiveness, in turn, and at least to some extent, is related to the fact that, in a democracy using the majority rule, a majority can exploit the outsider minority, for example by taxing it. In a parliamentary democracy this majority power is permanently anchored to the executive.

As regards the electoral laws used for appointing the members of the legislature it is enough for the purposes here to consider two main systems. First there is the system with single-member constituencies in which, in each constituency, the candidate who gets a plurality of votes, is elected. Then there is the system of multi-member constituencies in which the mandates are distributed to the parties in proportion to their votes.

The electoral system affects the parties in two ways. Firstly the plurality system has a tendency to reduce the number of parties, in the extreme to two parties, whereas there are no such reductive forces operating in the proportional system. Secondly, in contrast to the plurality system, the proportional system puts strong means for enhancing discipline, and thus for the creation of stable and cohesive parties, in the hands of the party leaderships. The main factor here is that the candidates for the legislature are largely dependent on the party leadership, both for nomination and for campaigning.

Now, by combining the methods for appointing the executive, and the electoral systems, we get four main types of democratic constitutions. This classification is useful, even if it is not exhaustive. In particular constitutions which in some way combine presidentialism and parliamentarism, such as the French one, do not fit into the pattern. Still, for the purposes here, it is enough to consider just two of the four combinations, namely parliamentary constitutions with proportionalism, and presidential constitutions with plurality.

Parliamentary constitutions with proportionalism. The main examples can be found in Western Europe, and one of them is Sweden. Usually there are some five to ten parties, since there are no strong forces reducing their number. A party may be substantially bigger than required by a contingent threshold rule. The parties are usually centralized, disciplined, stable and cohesive. Parliamentarism gives the incentives to discipline and proportionality provides the means. The executives are usually coalitions, and coalition-building and its conditions are therefore central aspects of politics. The coalitions are, at least to some extent, likely to be exploitive in the same way as the coalition in figure 2 above. The parties are likely to concentrate on their own narrow interests and tactical needs, and neglect issues considered important by citizens in general. There is no elected official with a responsibility for the common good, and thus characterized by some impartiality. All elected representatives primarily serve a particular party. Politics aiming at satisfying special interests is likely to drive out politics with more general ambitions, and the political process will to a large extent be driven by short-term, narrow interests?28 More than in any other constitutional context politics here is party politics. A society characterized by this kind of politics is often called a partyocracy.

Referendums in countries with this type of constitution, since they usually tend to weaken the parties, and since the parties usually have the power to decide about them, are thus not likely to be often used. Sometimes, however, they can be used by the parties in their own interest. They can be used as an excuse for avoiding commitment in regard to issues which, in the regular parliamentary process, would threaten the parties' integrity. We should thus not expect to find many referendums in this kind of political system, and when we find them they should be about issues which divide a party, or parties, or perhaps some governing coalition of parties. Furthermore, this way of using the referendum institute obviously requires that the parties control it, and popular initiatives are therefore not likely to be institutionalized. On the other hand it may be argued that, from the citizen's point of view, referendums, and particularly initiatives, are desirable for exactly these same reasons. The elements of direct democracy may check the tendencies towards partyocracy.

Presidential constitutions with plurality. The main example here is the US. There we find, as is to be expected, two large parties with weak internal discipline. There is no way to predict when referendums or initiatives are likely to be used. On the one hand they are not likely to damage the system in any way; on the other hand the system is not, in any respect, dependent on them. Referendums and initiatives may be used or not. When they are, however, the reason may simply be a wish to let the people decide. Motives based on power politics are not likely.

9. US experiences

We have now reached a point where we are likely to profit from learning about the experiences of direct democracy in real polities. The countries which seem most interesting are the US, Switzerland, and Italy. One reason for being interested in these countries is the frequency of the use of referendums and initiatives - Italy uses these instruments considerably more than most other countries, and the US and Switzerland much more than all other countries. Another reason is similarity to Sweden, or relevance for Sweden, and there Switzerland, and still more Italy, really deserve our attention.

Starting with the US it should first be noted that the federal constitution does not provide for any possibilities for direct democracy at all. Federal politics is altogether representative. All direct democracy in the US takes places in the states, but there it thrives all the more. The main rules are the following.

Table 2: Number of referendums and initiatives in the US in the decades of the 20th century - a) projected.29

In table 2 we see that the overall use of these institutes has varied considerably in the 20th century. In particular it is interesting that there is, in the last few decades, a strong tendency towards increased use. To judge from various current media reports that tendency goes on unbroken, and especially that is true for initiatives. State initiatives constitute an important, and expanding, part of US politics today. But even so these tendencies are not evenly spread out over the union. Direct democracy has its stronghold in the West.30 A more detailed picture of the distribution is given in table 3.

Let us first look at the issues raised by means of initiatives. Many of them are, in fact, of a quasi constitutional character, or, as it is sometimes put, about governance policy. I am thinking about restrictions on politicians' and legislatures' capacity for, for example, taxing and spending (State Tax and Expenditure Limitations, or just TELs). The most well-known among such initiatives is probably California's Proposition 13, which puts limits on property taxes, and also stipulates very restrictive rules for tax changes, and which was accepted. Another similar example is Nevada's banning of income taxes.31 "Other examples that may qualify as governance policies", writes C Tolbert, "include affirmative action, campaign finance reform, and criminal sentencing. Anti-affirmative action initiatives have qualified for the 1996 ballot in over ten states, including California. These measures would prohibit state governments and universities from considering race, ethnicity, or gender in awarding employment, contracts, or college admissions. Public regulation of the courts and judiciary via direct democracy is also evident in the adoption of 'three-strikes-you're-out' measures that require the legal system to sentence repeat convicted felons to life in prison. These measures have been adopted in over 30 states nationwide."32

Table 3: Number of referendums and initatives 1950-1992 in the ten US states topping the list.33

This great interest among the citizens in quasi constitutional issues is remarkable. Taking into account the very common pattern that governments determine what private citizens and organizations are allowed to do, this is rather the opposite. In these cases, the private citizens determine what the government is allowed to do, or what it should do.

Not surprisingly a lot of other issues are however raised as well. Just to mention a few examples it has thus been proposed that cock-fighting should be prohibited, that gambling rights should be extended on Indian tribal lands, and that billboards should be banned. In short, the initiatives can be about just anything.

Scholars studying initiatives have also been interested in their political tendencies. Do the results, for example, usually have a liberal, or a conservative, tendency? Or do they tend to go in any other interesting and observable direction? There are various results from this kind of studies, but they are usually quite inconclusive. Anyway, the most important question in this category may be whether the citizens, when handling the initiative, tend to engage in majoritarian tyranny, and thus to abuse minorities, for example homosexuals, immigrants, racial minorities, or welfare recipients.

This question is dealt with, at considerable length and detail, by Donovan and Bowler. Their general conclusion, after having studied initiatives in American states statistically, is that the fear is unfounded. "The latter part of the twentieth century", they write, "does provide numerous examples of initiatives targeting unpopular minorities ... We argue that it is wrong to point to these high-profile initiatives and conclude that policy outcomes from direct democracy are more abusive of minorities than are outcomes from legislatures. State and federal courts have struck down nearly every state initiative that critics cite as being abusive of minorities. In many cases, state voters are less likely to pass initiatives targeting minorities than they are to approve initiatives in general."34 Illustrating the unconvincing nature of this conclusion there are, however, also observers with opposing opinions. Magleby and Patterson thus tell us that the consultant "Kimball, who was in the business of gathering signatures for more than twenty years, no longer takes initiative clients. He is concerned that in the 1990s the process is increasingly used to target minorities: 'You know, we hit the Mexicans last year, it's blacks this year, it's going to be gays next year'."35 It is also worth noting that Donovan and Bowler, in their reassuring argument, rely at least as much on confidence in state and federal courts, as on the qualities of the initiative process itself.

Another hypothesis, which has attracted interest, is that initiatives tend to favor organized and moneyed interests. For the discussion of this concern it is essential to distinguish between the problem of getting a proposal to the ballot, and the problem thereafter of making it pass. Looking first at the problem of getting a proposal presented to the people, and thus an object for voting, there is no doubt about the matter. One study on California has, for instance, concluded that "any individual, corporation or organization with approximately $ 1 million to spend can now place any issue on the ballot".36 Furthermore, without financial or organizational resources it is, in the general case, very difficult, or impossible, to bring an issue to the ballot. The second step, getting the proposition passed is, however, another matter. In fact, most initiatives fall.37 Furthermore, propositions put on the ballot by a legislature for the citizens' approval, that is referendums, are less likely to fall than propositions put on the ballot by the citizens themselves.38 In this light the voters seem to take up a conservative outlook and to reflect on the consequences of their vote. The political scientist E Gerber has, for instance, found that opulent special-interest groups such as the oil and tobacco industries almost never succeed in pushing through self-serving initiatives. Sometimes they have, however, had greater success in blocking measures aimed against them. Thus the overall conclusion about moneyed interests is tentatively, that in the general case financial and organizational resources are necessary, but not sufficient, conditions for successful initiatives.

It is, in this context, interesting and important to note that a very considerable initiative industry has developed in the US. These consulting firms, which often also have a considerable legal competence, are specialized in signature collecting, wording of propositions, campaigning, and so on.

In the preceding section, while discussing the compatibility of direct democracy with various representative systems, I claimed about presidential constitutions with plurality that "referendums and initiatives may be used or not". The general US pattern, we now can say, corresponds well with this prediction, or rather lack of prediction. In spite of the far-reaching similarities between the federal constitution, and the constitutions of the states, there is no direct democracy at the federal level, and quite a lot of it at the state level. It could, however, quite conceivably, have been the other way round, or we could have had a lot of direct democracy at both levels, or at none of them. Anyway, things being as they are, it is a general observation that the effects of direct democracy tend to spread. There are several mechanisms involved here. An initiative which is successful in one state, tends to be repeated in other states. Furthermore, representatives will sometimes, in order to improve their own political status, initiate issues, which otherwise are likely to be initiated by the people. In other cases, on the contrary, representatives will abstain from initiating legislation, which is likely to be rejected by the people. The implication of all this is that the direct democracy in the states is important for the political agenda setting in the nation in its entirety.39

10. Swiss experiences

In Switzerland both referendums and initiatives are important at the national level, at the cantonal level, and in the communes, and there is a very long tradition of direct democracy. At the national level, the main rules are these:

These rules are quite similar to the ones in the US states and, together with the US ones, considerably more far-reaching than in any other country. It is therefore interesting at this stage to make a comparison of the Swiss and the US rules. First, then, it should be noted that both in Switzerland, and in some US states, it is possible for citizens to write complete proposals for new laws, and to bring them to the people for a decision (direct initiative). This is not possible anywhere else in the world.40 In federal Switzerland, however, as we have seen, the possibility is limited to changes of the constitution. It is not applicable to laws in general. A second important point is that both Switzerland and some US states provide for popular referendums. In both cases a group of citizens can thus demand that a law, passed by the representative organs, is presented to the people for a final acceptance or rejection.

The Swiss have used the popular referendum quite a lot to actually reject laws passed by the Federal Assembly. In particular that was true before 1959, when the historical settlement about a more or less permanent distribution of the seats in the Federal Council was reached (two seats each for the Radical Party, the Christian Democratic People's Party and the Socialist Party of Switzerland, and one seat for the Swiss People's party).41

The initiative, as we have seen, is confined to constitutional issues, and the concept of constitutionality has therefore, in reality, been given a very broad interpretation. Many issues have been deemed constitutional, which would not have been so in other countries, thus "allowing the Swiss Constitution to become a hodgepodge of fundamental law mixed with routine legislation".42 In Switzerland it has thus been proposed that the national army should be abolished (rejected), that the use and sale of marijuana, heroin and cocaine should be legalized (rejected), and that genetic engineering should be curbed. A proposal about more progressive taxation has been rejected.43 A proposal about price control has been accepted.44 Thus, in sum, the initiatives cover a very wide range of topics.

In later years, as table 4 shows, the frequency of initiatives has also increased substantially. This, by the way, is in line with the development in the US. The same is true for the tendency that most initiatives are rejected.

Table 4: Frequency of initiatives in Switzerland.45

In the same way as in the US the handling of initiatives involves a lot of money, and has stimulated the emergence of consultant firms and the like. "Not surprisingly", writes Kobach, "a cottage industry has grown around the initiative device, providing services ranging from signature collection to direct mailing to professional campaign management."46 With reference to Switzerland in particular it is however also important to note that interest organizations, when strong and developed enough, are perfectly able to handle the elements of direct democracy themselves - they need not apply to any consultants. Several authors have written about Switzerland from this point of view. Hughes says, for example, that "forming opinion is an expensive business, and the matter is too important to be left to luck and a chance majority. Swiss democracy is geared to pressure groups: it is a form of government calculated to call such groups into existence and give them power. The system could conceivably continue for a time without parties, but without pressure groups it would not work at all."47

So far we have registered some important similarities in the US and the Swiss experiences of direct democracy. There are however also some very important differences, which have to do with the effects on the political system at large. In the US there certainly were some such effects, but not at all, as we shall see, as thorough and fundamental as in Switzerland. Still, it is quite easy to neglect these effects in Switzerland altogether. It might be said that the country's very frequent use of the instruments of direct democracy need not be surprising, since it is a non-parliamentarian country, which is true. That argument, however, is much too superficial, and disregards crucial dynamic aspects. In particular it disregards the very important point that Switzerland has become non-parliamentary because of the old and very strong traditions of direct democracy. It is, for example, not the case that some constitutional engineers, thinking about the compatibility of the properties, have designed a constitution which is non-parliamentarian, and which provides for a lot of direct democracy. Rather, the rest of the constitutional machinery has been forced to adapt itself to a life in harmony with the old and strong elements of direct democracy.

It is not possible to elaborate these matters in detail here, but a few quotations may be used for giving an indication of the ideas. "Today", Kobach writes, "Switzerland is ... known for its consociational politics. Since the introduction of the 'magic formula' in 1959, the seven seats on the Federal Council have been distributed in a continuous 2:2:2:1 seat distribution among the largest parties ... This grand coalition, combined with a pervasive ethos of compromise, has prompted many an observer to label Swiss government a textbook case of Konkordanz-Demokratie."48 Then, going on to the causes, Kobach says that "this was not always the case, however... Consociational behaviour and institutions grew by degrees, beginning in the 1880s and culminating in the formula of 1959. At every stage, the referendum was the primary engine behind the change... It is difficult to overstate the importance of direct democracy in bringing about this facet of Swiss politics." In the same spirit Kobach also says that "direct democracy has contributed greatly to the relative weakness of political parties in Switzerland. It has rendered redundant an otherwise important party function, the aggregation and articulation of policy preferences in the electorate, by allowing citizens to express their opinions directly. At the same time, the initiative has undermined the parties' control of the national agenda. Referendums and initiatives have also weakened Swiss parties directly by undermining their unity."49

There are some empirical data which are very puzzling, but which, perhaps, may be explained by this unique Swiss relation between the elements of representative and direct democracy. I am thinking of the general election record. In table 5 we see that the electoral support for the main parties, over a long period of twenty-four years, is extraordinarily, and surprisingly, stable. What is the reason for this?

Kobach's explanation is in terms of the interaction between representative and direct democracy. "This combination of weak parties with strong voter loyalty is no mere coincidence", he writes. "The referendum affords voters the opportunity to dissent from their party's position without actually leaving the party. Because Swiss voters are accustomed to disagreeing with their party of choice on specific issues, party loyalty in elections tends to be based on factors more lasting than transient policy positions. Traditional family allegiances and general party images are more important. Thus, referendums have fostered looser, but more resilient party ties."50

Table 5: The results of Swiss general elections by pecentage of the vote. Abbreviations: FDP/PRD = Radical Party; CVP/PDC = Christian Democratic People's Party; SPS/PSS = Socialist Party of Switzerland; SVP/UDC = Swiss People's Party. The figures after the party-names stand for the number of seats in the Federal Council.51

It might be added that the stable voter support for the parties seems to be a prerequisite for the "magic" 2:2:2:1 formula, which, in turn, seems to be a prerequisite for the strong tendency towards consensus. "In sum", says Kobach, "the referendum and Swiss consociationalism have developed hand in hand. Consequently, Switzerland's Konkordanz-Demokratie is more firmly entrenched and more widely accepted than it might have been otherwise. Ironically, the blunt majoritarianism of the referendum has done much to foster the politics of consensus."52

11. Italian experiences

Even if the Swiss record is highly relevant for Sweden it may nevertheless be remarked that the processes described have taken a considerable time to develop. From that point of view Italy is more interesting, since similar observable changes have occurred in just a few decades. The constitution of 1947 made Italy a fairly clear parliamentary, proportional democracy and elements of direct democracy were therefore anathema - but they still existed. The reason was a will to repeal fascist ideas of strong leadership. In particular there was the so called abrogative referendum.

But even if the constitution of 1947 contained the basic rule for the abrogative referendum it was not used until after 1970, when a necessary implementing law was enacted as well. At that time the Christian Democrats were in a minority in the legislature and a divorce law had been taken against their will. Believing that they could reject that law in a referendum they therefore did something they had refused to do until then. They gave their support to the law which was needed for making the abrogative referendum operative. Then, when the referendum about the divorce law finally was held, in 1974, the law was upheld. The Christian Democrats and other Catholic groups, contrary to their own prediction, lost. And the cat, the ever present possibility of abrogative referendums, was out of the bag.54 After that the small Radical Party pioneered the systematic use of the referendum institute for promoting the secularization and liberalization of the Italian society. Gradually the institute has become more and more used, and by many different kinds of groups. The referendums have been about various issues including, especially, divorce and abortions.

Here I am mainly interested in the effects of direct democracy on the political system at large, and those effects are considerable. According to Bogdanor "the main effect of the divorce and abortion referendums was to reveal the Italian electorate as both more secular and more liberal than had been generally suspected. The outcome of the referendums threatened the hegemony of the Christian Democrats over the political system."55 An other referendum, says Bogdanor, "undermined the other main pillar of the Italian republic: the Communist Party". Summarizing the history of the Italian political system after the second world war Bogdanor says that "the system came to be dominated by the political parties who governed in their own interests rather than in the interests of those they claimed to represent. Italy had become a partiocrazia - a 'partyocracy'. As the parties lost their authority, with deference disappearing and a desire on the part of the electorate for greater participation, use of the referendum gathered pace. So the referendum is both cause and consequence of the transformation of the Italian political system."56

Again, as in the case of Switzerland, we thus see that if, for one reason or another, there are elements of direct democracy in a representative system, those elements are not necessarily ousted, even if not tolerated by the system. The result, rather, may be a fundamental transformation of the representative system - and the overturn of an unduly dominating political culture.

12. The case of Sweden - experiences

The present Swedish constitution provides for two types of referendums at the national level.

The first type, which concerns constitutional changes, is a complement to the normal procedure for such changes. According to the normal procedure a first decision in the legislature about a constitutional change has to be confirmed by a second decision, when the legislature has been reappointed in general elections. According to the alternative procedure it is, however, also possible to present the proposal for the people in a referendum after the first decision in the legislature, and that shall be done if demanded by at least a third of the members of the legislature. If so that referendum will take place at the same time as the next general elections. At this stage the proposal may be definitely rejected, but not definitely accepted. Thus, if it is not rejected, it will be handled according to the normal procedure by the new legislature, which means a second vote. For the proposal to be rejected two conditions must be fulfilled. There must be a majority against the proposal in the referendum, and those voting against the proposal in the referendum must exceed half the number of votes in the general election.

So far no constitutional referendum has been called. In at least one situation the institute has however been effective as a threat. In 1984 the Social Democratic government presented a proposal which expanded the voting rights for immigrants to Sweden, and similarly reduced those rights for Swedes abroad. The proposal was controversial, and the bourgeois parties threatened to refer the issue to a referendum. The threat made the government withdraw its proposal.57 This, by the way, corresponds well with the experiences reported in the last paragraph in the section about the US above.

The second type of referendum concerns other issues than constitutional ones. Here, the legislature, by a simple majority vote, may present any issue to the people. The idea, however, is not that the people should ratify a decision already taken by the legislature. The legislature is only, as the saying goes, asking the people for its advice. Referendums of this type are therefore not decisive, only consultative.

Until now (1999) Sweden has had five referendums.58 The first one, in 1922, was about prohibition of alcoholic beverages. 49 percent of the voters favored prohibition and 51 percent were against. The second one, in 1955, was about a change from left to right side driving. Only 17 percent of the voters wanted a change, 83 percent rejected it. Several years later, in 1963, it was nevertheless decided by the legislature that right side driving should be introduced, which happened in 1967. The third referendum, in 1957, was about the introduction, and design, of a supplementary pensions system. This referendum was remarkable since three positive alternatives were presented to the voters rather than, as usual, one positive alternative and its status quo preserving negation. The presence of three alternatives made the result of the referendum ambiguous and several interpretations were possible. Eventually the whole issue was settled by the government and the legislature. The debate in the legislature was intense and bitter and the vote very close, 115 votes in favor of the government's proposition and 114 against. The referendum institute, as it was used in this particular case, had hardly contributed to consensus, rather the opposite. The fourth referendum, about nuclear power, was held in 1980. Again there were three positive alternatives, and again the result was ambiguous. In spite of this a certain amount of consensus emerged and the following decision in the legislature, namely that all nuclear power plants should be closed by the year 2010, was not at all as controversial as the parliamentary decision following the referendum about supplementary pensions. However, the relative political peace was, as it now seems, illusory. At the same time as it has become gradually more evident that the decision will not be implemented, the political controversies have surfaced anew. The fifth referendum, in 1994, was about the Swedish membership of the EU. From a formal point of view this referendum, as the four other ones, was consultative. In reality, however, it came close to being decisive, since all the political parties in advance committed themselves to honoring the result of the referendum. 52 percent of the voters said yes to EU, and 47 percent said no. Immediately after that Sweden became a member of the EU.

When summarizing the experiences from these five referendums it may first be noted that three of the referendums have served as experiments of some kind, or full scale exercises, in periods when the referendum institute was intensely discussed and investigated by governmental committees. This is true for the referendum in 1922, and for the referendums in 1955 and -57. The events around the referendum in 1922 resulted in the institute of an optional, consultative referendum, initiated by the government, which Sweden has had since then. Furthermore it is interesting that none of the referendums have had the single purpose of finding out what people think, and only for one or two of them has that been a main purpose. In particular it is evident that a main reason for presenting the EU issue to the people in 1994 was the quest for a well-founded decision, and thus for legitimacy. In view of our earlier discussion about the special character of constitutional issues this was, of course, as it should be. The 1994 referendum is, however, a bit of an exception. The dominating purpose of the referendums has rather been to bypass problems of divided opinions within political parties, or, perhaps, within governmental coalitions. Problems of this kind, in good accordance with the discussion in the section "the constitution and the political parties", have been important in each of the five cases. The Swedish consultative variety of the referendum institute has also been a useful tool in these situations, permitting, for instance, the political manoeuvring resulting in three alternatives, and thus ambiguous outcomes, in the referendums of 1957 and 1980. Thus, and basically, the national referendum institute in Sweden has been a means in the hands of the parties, helping them to consolidate their own power positions or, in short, to bolster the Swedish partyocracy.

In addition to the national institutes there are also elements of direct democracy at the local community level. There is a possibility for consultative referendums, and there is also a weak indirect initiative.59

13. The case of Sweden - proposals for change

Even if the Swedish history of referendums so far is hardly encouraging, it need not necessarily be that way. There may be room for improvements. When discussing these things I shall concentrate on the national level, disregarding the local communities. I shall furthermore disregard the rare problem of constitutional change and rather concentrate on the use of referendums and initiatives in the regular, ordinary decision-making process concerning other laws than the constitutional ones. The examples of US, Switzerland and Italy indicate that such everyday use of direct democracy may, in various ways, be of considerable value.

Since Sweden is a parliamentary country with proportional elections the representative decision-making bodies are, according to the ideas presented above, likely to commit a lot of sins of commission, and a lot of sins of omission. The basis for this conjecture is given in the section "the constitution and the political parties" above, where it is said about countries with parliamentarism and proportionalism that "(p)olitics aiming at satisfying special interests is likely to drive out politics with more general ambitions". Here the politics satisfying special interests constitute the sins of commission, and the lack of politics with general ambitions the sins of omission. This, in turn, and following Bogdanor, clearly suggests significant tasks for referendums, and for initiatives. The referendums could be a corrective for the sins of commission, and the initiatives a corrective for the sins of omission. So let us consider a referendum institute, and an initiative institute, and discuss their arrangement in same detail.

Starting with the referendum institute it should, I think, be decisive, and the ballots should always be about decisions already taken by the representative bodies. The typical procedure would thus be that a decision taken by the legislature is presented to the people, acting as an extra house, for ratification, or rejection. The main reason for suggesting referendums of this kind, of course, is the intention to deal with sins of commission. In order to make that possible the sin really has to be committed, which means that there has to be a decision. Furthermore, and since the corrective is intended to be effective, the people's verdict obviously has to be decisive, or final.

The method may be varied in several ways.60 The presentation of the issue to the people may be mandatory or optional. Relying on international experiences, and in order not to exhaust the electorate unnecessarily, it seems preferable to have only optional referendums. Even so, however, the requisite for getting the issue to the people may vary. It may, for example, be required that the submission be demanded by a legislative minority exceeding a certain size, by the speaker in the legislature, or by a certain minimal number of citizens. There may be arguments in favor of the one or the other alternative. What all of the suggested initiators have in common, however, is that they are not the same as the government, or the legislative majority. Otherwise, of course, it wouldn't be a method for dealing with sins of commissions.

Taking into account the arguments in the section about non-voting above it may be prudent to introduce a quorum stipulation. For the same reason it may also be good to execute all impending referendums at one single occasion, for example in connection with general elections. The purpose of having only optional referendums, and no mandatory ones, as suggested above, is of course also to keep the turnouts high.

Let us then go to the sins of omission. Here, as I have already mentioned several times, the initiative may be considered a proper corrective, and the initiative is then taken to be a popular initiative, exactly as in the definition on the very first page of this paper. In a country like Sweden there are however, in my opinion, considerable drawbacks associated with an initiative of this particular kind, and the question, whether there is an alternative, therefore arises.

The drawbacks, to start with them, are easily spelled out. We have already seen that an extensive consulting industry has developed in the US in order to help people to undertake initiatives. In Switzerland there is a similar development, but the direct democracy has also developed in symbiosis with the traditional interest organizations. Those organizations are, of course, in excellent positions for collecting signatures, for promoting various causes, or, in short, for using an initiative institute for their own purposes. In a country where the organizations are already strong, an ordinary initiative institute is therefore likely to make those organizations still stronger. Now, as I see it, one of Sweden's main problems is that the organizations are already too strong.61 The introduction of a popular initiative would therefore be ill advised. But, perhaps, there is some kind of alternative?

Above, when considering referendums, I discussed situations in which the people, in a referendum, is given the power to defeat a proposition which has been passed by the legislature. Logically there is, however, also the possibility of giving the people the power to affirm a proposition which has been rejected by the legislature. The distinction between these two cases is important, and the second possibility may, in fact, be the kind of substitute for a proper initiative that we are looking for. Assume, for example, that a legislative minority, exceeding a certain size, can demand that a proposition, which is rejected in the legislature, be put to the people. If such a group of legislators believe that a certain proposition, which would be stillborn in the legislature, will gain the support of a majority of the citizens, they can raise the issue in the legislature anyway. Then, after the proposition is defeated there, they can get it presented for the people. An arrangement of this kind can, I think, serve as an important and effective substitute for an ordinary initiative institute.

It is, for example, easy to imagine that a small party group outside the executive coalition may take initiatives of the kind mentioned. If so the group acts as a rather homogenous entity, or as a single actor, in the way that is common in parliamentary politics. In a longer run it is however also possible that the elements of direct democracy which we are discussing contribute to the undermining of party discipline, in particular since there is in Sweden by now a possibility for the voters, in the general elections, to complement their support for a party with support for particular candidates. Thus there are incentives for individual members of the legislature to cultivate their own profiles and images, and legislative initiatives may be an important means towards that end. A legislator seeking a new appointment may, for example, in his campaign refer to his participation, in an ad hoc group, in the initiation of a successful and generally appreciated proposal, which has been accepted by the people. Such behavior presupposes that the party discipline is not too rigorous, and it also contributes to loosening the party discipline further. I therefore believe that virtually all kinds of proposals, which are likely to be accepted by the people, may also materialize in the way discussed here. It may, for example, be proposals about an upper tax limit, about a change to plurality elections in single-member constituencies (first past the post), about requiring qualified majorities for some kinds of decisions in the legislature. But if these reforms are not likely to emerge without the prior introduction of the elements of direct democracy suggested here, then the same may be true for these elements themselves. In Italy, as we saw, the cat did not get out of the bag until the Christian Democrats made a fateful mistake. Perhaps the problem will get some kind of solution in Sweden as well?

At this point, and before finishing, it may be a good idea to compare shortly the proposals for change made here with those made by a recent governmental committee on direct democracy (Folkomröstningsutredningen).62 Thus the committee proposes the possibility of both consultative and decisive referendums, whereas I have no place for the consultative ones. As for the decisive referendums my proposal is that their objects should be decisions already taken in the legislature (or perhaps by some other representative body), while the committee has a similar alternative (as well as another one). The basic difference between my proposal and that of the committee concerns the rules for calling a referendum. The committee holds that only a majority in the legislature should be able to call a referendum. I, on the contrary, suggest several possibilities, except a legislative majority, which is ruled out. The implication of the committee's position here, as I see it, is that no corrective for sins of commission is introduced. Finally, the committee does not propose an initiative, or an alternative to such an institute, whereas I do. The committee thus has no corrective for sins of omission.

Now, going back to my own argument, the reason, so far, for proposing a new referendum institute, and an institute substituting initiatives, has been a wish to correct sins of commission, and sins of omission. In my view there is however also another, more important, and more fundamental, reason for the suggested changes. As the examples of Switzerland, and in particular Italy, show, elements of direct democracy may be instrumental in transforming a representative system with strong and dominating parties. Sweden, according to my opinion, is a typical partyocracy. The political life is dominated by the political parties, and in particular by their leaderships. Furthermore the Social Democratic Party is the center of a dominating political culture in about the same way as the Christian Democrats in Italy. It would be a good thing, I contend, if these structures were weakened. In the same way as in Italy, the suggested elements of direct democracy, may work in that direction. That possibility, I should say, is my main reason for proposing the introduction of new, and sharper, instruments of direct democracy into the Swedish political system.

The suggested changes would thus, in all likelihood, make the political parties less coherent, centralized and disciplined. This would also, almost automatically, make interest organizations and unions weaker, since they would lose their counterparts at the tops of the political hierarchies. Even if these changes are intended they may, however, also entail some problems. In particular they would impair the functioning of the parliamentary system. The remedy for this could, for example, be a permanent broad executive of the Swiss type, or some kind of (semi)presidentialism. This, though, is not the place for discussing the implications of such changes. I will just end by saying that, at least for me, they are welcome.


1There are several works giving general descriptions and analyses of referendums and initiatives. In particular I would like to mention Butler & Ranney, 1994; Gallagher & Uleri, 1996; Suksu, 1993; Bowler, Donovan & Tolbert, 1998; and SOU 1997:56.
2Butler & Ranney, 1994, p 4
3Source: LeDuc, Niemi & Norris, 1996. In order to include the EU referendum in 1994 the figure 3 for Sweden in the source has been changed to 4.
4Butler & Ranney, 1994, p 6.
5Brady & Kaplan, 1994, p 175.
6Bogdanor, 1994, p 36.
7Rather detailed descriptions of these referendums are given in Brady & Kaplan, 1994, and White & Hill, 1996.
8The Pareto- and Kaldor-Hicks-criteria, which are used here, are described and explained in a lot of elementary textbooks in Economics. Moberg, 1998 b, chapter 4, has a discussion about the problems involved in applications to political decisions.
9This ideal is not universally acknowledged. Butler & Ranney, 1994, p 11f, for instance, write that "While political theorists and practitioners have long used the term democracy in many different ways, most appear to agree that it denotes at least the principles of popular sovereignty, political equality, popular consultation, and majority rule." This is sharply at odds with this paper's philosophy. Here, following among others Wicksell, 1896, and Buchanan & Tullock, 1962, unanimity is the principle and the majority rule only a necessary adaption to pradical realities. It is indeed difficult to understand how the majority rule, considered as a principle, can be compatible with the principles of popular sovereignty, political equality and popular consultation. Taken as a principle the majority rule rather makes a mess of popular sovereignty, political equality and popular consultation.
10Manin, 1997, contains an interesting discussion about the ideas of representation, and their historical development.
11Here it is appropriate to recall the distinction made by James Buchanan between politics as an exchange process, and politics as truth finding. Buchanan has emphasised (Buchanan, 1992, p 74) "the conceptualization of politics as exchange". In another context, together with Geoffrey Brennan, he contrasts this contractarian approach to the scientific approach. They write (Brennan & Buchanan, 1985, p 38) that "The contractarian contends that politics is best modeled as being analogous to the economic relationship; the noncontractarian contends that politics is best modeled as being analogous to scientific activity". Obviously Buchanan discards the idea of politics as a scientific truth finding process. The idea of a representing expert élite is, however, close to the idea of politics as truth finding. Therefore, if we give some room for experts, we also have to admit that politics, at least partially, consists in, or aims at, truth finding.
12Bogdanor, 1991.
13Buchanan & Tullock, 1962.
15Hirschman, 1970, p 76.
16Niskanen, 1971, p 19.
17The German reunification, however, did not involve any referendum.
18Bogdanor, 1994, p 58.
19For further details see Wallin, 1997.
20Butler & Ranney, 1994, p 10.
21Butler & Ranney, 1994, p 8.
22Bogdanor, 1994, p 34ff.
23In a referendum in April 1999 a complete elimination of proportionalism failed, since the turnout of 49,6% was below the 50% required by a quorum rule.
24Demsetz, 1990.
25Gilljam, 1995. This is the original Swedish text: "Det är svårt att se varför en folkomröstning med ett valdeltagande på 75 procent kan vara ett bättre uttryck för folkviljan än ett parlamentsbeslut i samma fråga fattat av folkvalda representanter valda med 90-procentigt valdeltagande."
26Bogdanor, 1991.
27This topic is elaborated in detail in Moberg, 1998 b, and to some extent also in Moberg, 1998 a.
28This particular point is elaborated in Moberg 1998 a, p 61.
29Magleby, 1994, p 232.
30The political ideas characteristic for the Progressive Era, ca 1890-1920, favoured popular power by means of referendums, initiatives, recall of legislators, primaries, direct election of senators, and so forth. These ideas became particularly important in the West, both because they were most widely hold there, and because those parts of the nation were still in the early stages of their political development at that time. (Magleby, 1994, p 219, 223)
31Magleby, 1994, p 228.
32Bowler, Donovan & Tolbert, 1998, p 175.
33Magleby, 1994, p 226.
34Bowler, Donovan & Tolbert, 1998, p 17.
35Magleby & Patterson, 1998, p 168.
36Bowler, Donovan & Tolbert, 1998, p 35.
37Bowler, Donovan & Tolbert, 1998, p 191.
38Butler & Ranney, 1994, p 21; Bowler, Donovan & Tolbert, 1998, p 121.
39Magleby, 1994, p 218.
40Butler & Ranney, 1994, p 19.
41Kobach, 1994, p 109; p 132.
42Kobach, 1994, p 106 f.
43Kobach, 1994, p 142.
44Kobach, 1994, p 145.
45Source: Kobach, 1994, p 141.
46Kobach, 1994, p 107.
47Hughes, 1962, p 34.
48Kobach, 1994, p 101.
49Kobach, 1994, p 132.
50Kobach, 1994, p 133.
51Source: Steinberg, 1996, p 117.
52Kobach, 1994, p 151.
53Bogdanor, 1994, p 62.
54Bogdanor, 1994, p 65.
55Bogdanor, 1994, p 66.
56Bogdanor, 1994, p 69.
57SOU 1997:56, p 39.
58Ruin, 1996, gives a detailed presentation of the Swedish referendum history.
59For further details see for example Wallin, 1997.
60In the public Swedish debate the former vice speaker of the legistature, Bertil Fiskesjö, has, for one, argued in favour of methods similar to the one discussed here. Fiskesjö, 1994 a, b.
61This point is elaborated in Moberg, 1999.
62SOU 1997:56.


Bogdanor, V, 1991, The entry "referendum and initiative", in The Blackwell Encyclopaedia of Political Science, (ed V Bogdanor). Oxford: Basil Blackwell Ltd, 1991.

-, 1994, "Western Europe", in Butler & Ranney, 1994.

Bowler, S, Donovan, T & Tolbert C J (ed), 1998, Citizens as Legislators, Direct Democracy in the United States. Columbus: Ohio State University Press, 1998.

Brady; H E & Kaplan, C S, 1994, "Eastern Europe and the Former Soviet Union", in Butler & Ranney, 1994.

Brennan, G & Buchanan, J M, 1985, The Reason of Rules, Constitutional Political Economy. Cambridge: Cambridge University Press, 1985.

Buchanan, J M, 1992, Better than Plowing and other Personal Essays. Chicago: The University of Chicago Press, 1992.

Buchanan, J M & Tullock, G, 1962, The Calculus of Consent, Logical Foundations of Constitutional Democracy. Ann Arbor: The University of Michigan Press, 1962.

Butler, D & Ranney, A (ed), 1994, Referendums Around the World - The Growing Use of Direct Democracy. London: Macmillan, 1994.

-, 1994, The chapters "Practice", "Theory", and "Conclusion", in Butler & Ranney, 1994.

Demsetz, H, 1990, "Amenity Potential, Indivisibilities, and Political Competition", in Perspectives on Positive Political Economy, (ed J E Alt & K A Shepsle). Cambridge: Cambridge University Press, 1990.

Fiskesjö, B, 1994 a, "Rådgivande folkomröstningar har stora brister" , in Svenska Dagbladet , March 29, 1994.

-, 1994 b, "Nyordning för folkomröstning krävs", in Svenska Dagbladet, November 17, 1994.

Gallagher, M & Uleri, P V (ed), 1996, The Referendum Experience in Europe. London: Macmillan Press Ltd, 1996.

Gilljam, M, 1995, "I valet mellan cynism och hederlighet", in Svenska Dagbladet, May 21, 1995.

Hirschman, A O, 1970, Exit, Voice and Loyalty -Responses to Decline in Firms, Organizations, and States. Cambridge (MA): Harvard University Press, 1970.

Hughes, C, 1962, The Parliament of Switzerland. London: Cassel & Company Ltd, 1962.

Hughes, C A, 1994, "Australia and New Zealand", in Butler & Ranney, 1994.

Kobach, K W, 1994, "Switzerland" in Butler& Ranney; 1994.

LeDuc, L, Niemi, R G & Norris, P, 1996, "Introduction - The Present and Future of Democratic Elections", in Comparing Democracies, Elections and Voting in Global Perspective, (ed L LeDuc, R G Niemi & P Norris). Thousand Oaks: SAGE Publications, 1996.

Magleby, D B, 1994, "Direct Legislation in the American States", in Butler& Ranney, 1994.

Magleby, D B & Patterson, K D, 1998, "Consultants and Direct Democracy", in Political Science & Politics, 31 (June 1998).

Manin, B, 1997, The Principles of Representative Government. Cambridge: Cambridge University Press, 1997.

Moberg, E, 1998 a, "The Expanding Public Sector - A Threat to Democracy?", in The Limits of Government - On Policy Competence and Economic Growth, (ed G Eliasson & N Karlson). Stockholm: City University Press, 1998.

-, 1998 b, Democracy - Constitutions, Politics and Welfare Effects, unpublished book manuscript.

-, 1999, "The Swedish Model - A Comment on Mancur Olson's Analysis", in A Not-so-Dismal Science - A Broader View of Economies and Societies, (ed M Olson & S Kähkönen). Oxford: Oxford University Press, 1999.

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

Ruin, O, 1996, "Sweden - The Referendum as an Instrument for Defusing Political Issues", in Gallagher & Uleri, 1996.

SOU 1997:56, Folket som rådgivare och beslutsfattare, Betänkande avgivet av Folkomröstningsutredningen. Stockholm: Fritzes,1997.

Steinberg, J, 1996, Why Switzerland?. Cambridge: Cambridge University Press, 1996.

Suksi, M, 1993, Bringing in the People - A Comparison of Constitutional Forms and Practices of the Referendum. Dordrecht: Martinus Nijhoff Publishers, 1993.

Tiebout, C M, 1956, "A Pure Theory of Local Expenditure", in Journal of Political Economy, 64 (October 1956).

Uleri, P V, 1996 a, "Introduction", in Gallagher & Uleri, 1996.

-, 1996 b, "Italy - Referendums and Initiatives from the Origins to the Crisis of a Democratic Regime", in Gallagher & Uleri, 1996.

Wallin, G, 1997, "Det kommunala folkomröstningsinstitutet", in Folkomröstningar och demokrati, SOU 1997:56, Bilaga 2. Stockholm: 1997.

White, S & Hill, R J, 1996, "Russia, the Former Soviet Union and Eastern Europe - The Referendum as a Flexible Political Instrument", in Gallagher & Uleri, 1996.

Wicksell, K, 1896, "Ein neues Prinzip der gerechten Besteurung, Finanztheoretische Untersuchungen", translated to English and reprinted in Classics in the Theory of Public Finance, (ed R A Musgrave & A T Peacock). London: Macmillan, 1967. Published in Swedish: Om en ny princip för rättvis beskattning (preface E Dahmén, transl L Björk). Stockholm: Ratio, 1987.