I think that it is appropriate that I devote this Johan Skytte Prize Lecture to a general subject in political science, and what I should like to do is to present some reflections on alternative and competing approaches in our discipline: cultural, structural, functional, and rational-choice (or publicchoice) approaches. I shall do so in the context of my research on consensus and consensus democracy - the work that the Johan Skytte Prize Committee cited as the contribution that merited the award.

Two different approaches, or at least two different emphases, are already apparent in the terms consensus and consensus democracy. Consensus means "group solidarity in sentiment and belief or "general agreement"; I am citing Webster's dictionary definition which I think is in accord with political science use of the term as well, although political scientists are, of course, especially interested in political consensus. This means that consensus is a cultural concept, since political culture can be defined - paraphrasing Gabriel A. Almond and Sidney Verba (1963, 14-15) - as a pattern of cognitive, affective, and evaluational orientations toward political objects among the

Resumé

Five examples of the origin, operation, and consequences of consensus democracy and closely related democratic forms (the politics of accommodation, consociational democracy, and power-sharing democracy) illustrate the relative strengths of cultural, structural, functional, and rational-choice explanations. The examples show that each of these explanations plays a crucial role in at least one of the five situations. Hence they suggest that it is unwise to assume that one particular approach predominates to such an extent that it should be the ruling paradigm for political research.

---

*Arend Lijphart, Department of Political Science, University of California, San Diego, 9500 Gilman Drive, La Jolla, California 92093-0521, USA.

members of a group, or, in simpler words, a group's pattern of political beliefs, feelings, and judgments. However, consensus democracy, as I have used the term, refers not to culture but to structure: it is a pattern of institutional characteristics that particular democracies have, that is, their types of executives, legislatures, party systems, and so on.

Consensus Democracy and Related Terms

In fact, my term has been faulted by critics who argue that my consensus democracy is found in countries that do not necessarily have a high degree of consensus. I agree with this as a factual observation. However, I still believe that the term consensus democracy makes sense, because this kind of democracy can be seen as an institutional arrangement that is able to produce as much consensus as possible in countries, such as ethnically and religiously divided societies, where a spontaneous consensus is in short supply.

Perhaps it would be wise to search for a different term, but I should point out that "consensus democracy" is already my fourth attempt to find a term that can express this concept without causing misunderstandings or other objections. My first term was the "politics of accommodation," which I used in my case study of politics in the religiously and culturally divided Netherlands, my native country (Lijphart 1968). Here the criticism was that the term was not sufficiently distinctive because, it was argued, accommodation or compromise was the essence of politics generally and not characteristic of some systems and not of others. In a review article in the British Journal of Political Science, Brian Barry (1975, 477) also jokingly states that he has "come across a number of people who are firmly convinced that Arend Lijphart's book The Politics of Accommodation is a study of Dutch housing policy"!

In order to accommodate - if I may use this word just once more! - my critics, I switched to the term "consociational democracy" (Lijphart 1969; 1977). Consociation, or rather the Latin consociatio, was first used as a political concept in 1603 by the German political theorist Johannes Althaus or - to use the Latinized form of his name - Althusius, who lived at about the same time (1557-1638) as Johan Skytte (1577-1645). The great advantage of this new term to me was that it was no longer in common or academic use in the 20th century, so that my readers would be forced to listen to my definition without being distracted by their own preconceived notions. But the critics were still not happy: now they found my term too long, too polysyllabic, and too much of a tongue-twister. My next attempt at placating them was to use the simpler Anglo-Saxon term "power-sharing" as a synonym for consociation - for instance, in the title of my book Power-Sharing in South Africa (Lijphart 1985).

The final shift to "consensus democracy" cannot be blamed on my critics. What happened was that when I tried to systematically apply the concept of power-sharing democracy and its contrast with majoritarian democracy to a large number of democratic systems, I found it necessary to define these terms somewhat differently than I had done before. For reasons of clarity, I decided that a slightly different term was needed, too: hence the birth of "consensus democracy" (Lijphart 1984)! I think that I shall be stubborn and stick with this term - especially now that it has received the blessing of the Johan Skytte Prize Committee!

My research has focused to a large extent on political institutions, and I may be called an "institutionalist". But does this mean that I am exclusively committed to institutional or structural explanations? Let me use five major findings from my research on consociational and consensus democracy to show that this is not the case: I should like to tell five political science stories and the theoretical moral of each of these stories.

The Establishment of Power-Sharing in Deeply Divided Societies

My first story concerns the crucial decisions by political leaders to establish power-sharing in some of the deeply divided societies on which my work has focused: in Canada in 1840, in the Netherlands in 1917, both in Lebanon and in Switzerland in 1943, in Austria in 1945, in Malaysia in 1955, in Colombia in 1958, in Cyprus in 1960, in Belgium in 1970, in Czecho-Slovakia in 1989, and in South Africa in 1994. Four aspects of these decisions are worth highlighting. First, most of these decisions were made in situations of great tension and of potential or even actual violence. Second, the power-sharing systems that were set up all followed the same general pattern: an inclusive government consisting of representatives of all of the important rival groups; as much autonomy for these groups as possible; proportionality in representation and appointments; and a formal or informal minority veto power with regard to the most vital and fundamental matters.

Third, these decisions were made in different parts of the world and at widely different times - the countries that I have listed are located in five different continents, and there is more than a century and a half between the first and the last case I mentioned. Finally, these decisions were made completely independently of each other. With the exception of South Africa, none of the power-sharing agreements were inspired by the example of an earlier agreement of this kind; each time, power-sharing was re-invented. For instance, in 1958 the Colombian peacemakers were totally ignorant of, and hence could not learn any lessons from, the so-called Peaceful Settlement in the Netherlands in 1917 or the 1943 Lebanese National Pact.

What light does this story throw on the basic question that I raised? I would submit that the widely different times and places of these decisions to institute power-sharing rule out any cultural or structural explanations. Instead, powersharing was invented and re-invented time and again because of its compelling logic: it was the most rational choice to be made in the circumstances of potential or actual civil strife. The moral is that pure rational-choice decisions can and do occur. In fact, I know no other empirical example of rational choice that is as clear and convincing as this example and not one that concerns such momentous decisions.

Let me add a brief footnote to this story. The first modern scholar to identify the power-sharing model of democracy was not a political scientist but an economist: Sir Arthur Lewis. Lewis was born in St. Lucia in the Caribbean and was of African descent. He served as an economic adviser to several of the governments of West Africa from 1953 to 1965, and he observed and deplored the breakdown of democracy that was occurring in these countries. His diagnosis of this failure was that the West African ethnically divided countries had not been given the right kind of democracy. What they needed, he argued, was broad inter-ethnic coalitions, elections by proportional representation, and ethnic group autonomy. He did not attach a comprehensive label to these proposals, but they clearly added up to powersharing. He did not mention any empirical examples of power-sharing either, and he appears not to have known of the Colombian, Lebanese, Dutch, and other precedents. Hence, in contrast to political scientists like Gerhard Lehmbruch, Jiirg Steiner, Luc Huyse, and myself who discovered powersharing a few years later, Lewis invented power-sharing by trying to think what would be the logical solution to the problems in West Africa. This is another case of creative invention and of rational choice.

Lewis' analysis and recommendations were contained in a short book entitled Politics in West Africa (1965) - an outstanding book that, in my opinion, deserves a big award like the Johan Skytte Prize. But it was published in 1965, long before the establishment of the Skytte Prize, and Lewis died in 1991. Fortunately, Lewis won another major prize which is arguably at the same level of importance as the Skytte Prize: the 1979 Nobel Prize in Economic. He won it for his work in international economics.

The Clustering of Consensual Characteristics

My second story is taken from my systematic comparative analysis of all of the world's stable (durable) democracies: a total of 21 in my 1984 book Democracies, and a total of 36 in the updated and expanded version of this book that I am currently preparing. What I do in this study is to analyze the most important institutional characteristics of these democracies, and what I

find is that these characteristics are not randomly distributed but form two
distinct clusters.

One of these clusters consists of the types of executive power, executivelegislative relations, party system, electoral system, and interest group system. Democracies that have either broad governing coalitions or minority cabinets that are dependent on the shifting support of legislative majorities (instead of one-party majority governments) also tend to have relatively strong legislatures (instead of dominant, and even domineering, executives), multiparty systems (instead of two-party systems), proportional election systems (instead of plurality or majority electoral systems), and corporatist or coordinated interest group systems (instead of free-for-all competitive pluralism among interest groups). These are the democracies that I call consensus democracies, and the democracies characterized by the contrasting cluster of traits are the majoritarian democracies.

What is the moral of this story? Why is it that these five characteristics tend to go together? Here the moral is more complex than in my first story. Part of the explanation is clearly structural. Elections by proportional representation allow or encourage multiple parties to form and to gain representation in parliaments, and multiparty systems make it more likely that either coalition or minority cabinets will be formed.

But there is no such structural connection between multiparty systems and corporatist interest group systems. Joseph LaPalombara offers a functional explanation for the absence of such a link in the case of Italy - a democracy that I classify as consensual in most respects but that has a pluralist rather than a corporatist interest group system. LaPalombara (1987, 213-22) describes Italy as a partitocrazia with broad participation of all parties in policy-making and a strong inclination to seek consensus. The party leaders, he says, have a strong "aversion to divisive confrontations." This system is also called consociativismo in Italy - showing, among other things, that Italians do not have any problems with words consisting of many syllables! This description, of course, applies to the Italian system before the major reforms of 1994, which were aimed at moving Italian politics in the direction of winner-takeall majoritarianism and that made Gianfranco Fini, the neo-Fascist leader, happily proclaim: "Enough with consociativismo in management, in government" (cited in Cowell 1994). LaPalombara argues that the consensus produced by partitocrazia was so strong that there was simply no need for any further consensus to be produced by corporatism, and that the two should even be regarded as incompatible.

LaPalombara's functional argument seems persuasive: broad political coalitions and interest group corporatism can plausibly be seen as alternative methods of achieving consensus. In this view, weakness in interest group coordination may be compensated for by strong inter-party cooperation. This appears to be the case in Italy, but it is clearly not a general pattern in most

democracies; if it were, we would find a negative relationship between the other characteristics of consensus democracy and corporatism, but in faet there is a strong positive relationship. My nominee for a more plausible explanation is the cultural explanation. Consensus democracy and majoritarian democracy are alternative sets of political institutions, but more than that: they also represent what John D. Huber and G. Bingham Powell (1994) have called alternative "visions" of democracy. This is another way of saying that underlying the divergent types of institutions are divergent political cultures, a culture of consensus versus a culture of competition. This can account better for the general link between consensus characteristics and interest group corporatism than any structural or functional explanation.

The Clustering of Federal Characteristics

My third story has to do with the other cluster of democratic characteristics that I found in my 1984 book Democracies. Here the contrast is between federal and decentralized systems on the one hand and unitary and centralized systems on the other - a variation on the consensus-versus-majoritarian theme. Democracies that are federal and decentralized also tend to have bicameral legislatures with strong second chambers that are quite different in their composition from the first chambers (instead of unicameral or weakly bicameral legislatures), constitutions that are difficult to amend (instead of flexible constitutions), and constitutions that are protected by judicial review (instead of systems in which the courts do not have, or normally do not use, the power to review the constitutionality of legislation). What explains this pattern?

I must admit that, when I first discovered this pattern, I was puzzled by it. But I quickly realized that as a "new institutionalist" I had a lesson or two to learn from the "old institutionalists," especially theorists of federalism like K. C. Wheare (1946) and Carl J. Friedrich (1968). These theorists maintain that federalism has primary and secondary meanings. Its primary definition is: a guaranteed division of power between the central government and regional governments. Secondary characteristics are strong bicameralism, a rigid constitution, and strong judicial review. Their argument is that the guarantee of a federal division of power can only work well if both the guarantee and the exact lines of the division of power are clearly stated in the constitution and if this guarantee cannot be changed unilaterally at either the central or regional level (hence the need for a written and rigid constitution), if there is a neutral arbiter which can resolve conflicts concerning the division of power between the two levels of government (hence the need for judicial review), and if there is a federal chamber in the national legislature in which the regions have strong representation. This argument is a functional one: in order for

federalism to operate successfully, certain functions need to be performed,
and structures need to be created or adapted to perform these functions.

Consensus Democracies and Their Consequences

My fourth story is based on my recent work in which I have started to take a look at the "so what?" question: does the type of democracy make a difference for how well democracies perform? My first stimulus to do so was the view of many of my colleagues that consensus democracy might have many desirable qualities but that its coalitional and consultative style of decision-making was bound to be slow and therefore inefficient. My response was that, if this were so, the results should be apparent when we examine the performance of governments with regard to major policy areas like macroeconomic management - economic growth, inflation, and unemployment - and also with regard to the maintenance of public order and the control of violence. My empirical findings are that there are no significant differences: over-all, consensus democracies perform about as well as majoritarian democracies (Lijphart 1994).

I have also looked at the performance of the different types of democracy with regard to various qualities that are generally regarded as democratic desiderata, such as the representation of minorities and women and a high level of voter participation in elections. Here I find that consensus democracies clearly outperform the majoritarian democracies. The reason for this is partly structural, because consensus democracies generally use proportional representation (PR) as their electoral systems. PR makes it much easier for minorities and women to be elected, and PR also boosts voter turnout in two ways: by minimizing the wasted-vote problem, it makes it more attractive for voters to vote, and by making it more attractive for parties to campaign in areas where they are relatively weak, such stronger party efforts will also stimulate turnout.

However, an alternative or additional explanation would be cultural: consensus democracy itself and what I have just called its "consequences" may both be argued to spring from a general cultural inclination toward a strong community orientation and social consciousness. Another way to express this idea is to borrow former U.S. President George Bush's words. In his acceptance speech to the Republican national convention in 1988, Bush stated that he wanted "a kinder, and gentler nation." If there is such a kinder and gentler cultural predisposition, we should be able to find evidence of it in the policies pursued by consensus democracies. I have tested this in four policy areas: the welfare state, the environment, criminal justice, and foreign aid. And indeed, the consensus democracies are the kinder and gentler democracies: they are more likely than majoritarian democracies to be

welfare states, to be protective of the environment, to have less punitive criminal justice systems (as measured by their rates of incarceration and use of the death penalty), and to be more generous with foreign aid. The moral of this story is another victory for cultural explanation.

Continental European Versus Anglo-American Systems

My final story is based on the clustering not of institutional or policy characteristics, as in my previous stories, but on the clustering of countries. The contrast between consensus and majoritarian democracy is not a dichotomy but a continuum from strong consensus on one end to strong majoritarianism on the other. Particular democracies can therefore be more or less consensual or more or less majoritarian. But we can produce a dichotomous division by drawing a line in the middle, with the more consensual democracies on one side and the more majoritarian ones on the other. When we look at the Western democracies, we find, on the consensus side, Sweden as well as the four other Nordic countries, the Netherlands, Belgium, Luxembourg, Germany, Switzerland, Austria, Italy, and Portugal. On the majoritarian side we find the United Kingdom, the United States, Canada, Australia, and New Zealand.

This obviously looks very much like a division between Continental European countries on the one hand and Anglo-American countries on the other. It is reminiscent of the first modern attempt to systematically classify democracies: the 1956 article "Comparative Political Systems" by Gabriel A. Almond (who, incidentally, served as my dissertation supervisor at Yale, and whose work was my first inspiration to systematically examine different types of democracies). Almond's criteria were structural and cultural, but he used the geographical labels "Anglo-American" and "Continental European" to describe his two basic types. There are some exceptions to the neat twofold split: Ireland is almost exactly on the dividing line, and France, Spain, and Greece are on the "Anglo-American" side. It is also interesting to note, especially in the light of President Charles de Gaulle's fulminations against les anglo-saxons, that of all of the democracies physically located on the European continent the French Fifth Republic - often called "de Gaulle's republic" (Williams 1961) - is the most majoritarian, that is, the most Anglo To my mind, this geographical pattern strengthens the case for cultural explanation; I think that there is indeed, as Almond suggested more than forty years ago, a major difference between the political cultures of the Anglo-American and Continental European worlds, and that this cultural difference manifests itself inter alia in the different forms of democracy that are practiced in these two worlds.

Conclusions

My five stories clearly have different morals. I hope that I have been able to show that cultural, structural, functional, and rational-choice explanations can all be persuasive and legitimate for different phenomena or situations to be explained. While it is also perfectly legitimate to formulate hypotheses based on just one of these approaches, I believe that it is unwise and self-defeating to focus on one to the complete exclusion of the others. If examples of cultural explanations have predominated in my stories, as I think they have, this is probably mainly due to chance; my five stories are obviously not a representative sample of the thousands of political science stories that could be told.

Moreover, I believe that especially political culture and political structure tend to interact very closely with each other. As Almond & Verba (1963, 35) already argued in The Civic Culture in 1963, structural and cultural phenomena are variables in "a complex, multidirectional system of causality." My conclusions about consensus and consensus democracy are that the structure of consensus democracy may either be based on a consensual culture, or that it may operate in an insufficiently consensual culture in such a way as to first produce the minimum of consensus required for a democracy and then, in the long run, make the country's political culture more consensual. That is, the structure of consensus democracy may be the product of a consensual culture or its causal agent. When the latter happens, of course, the country in question can also afford to move from a consensus system of democracy to a more majoritarian one. But that is not what I would necessarily recommend: why give up consensus democracy and its "kinder, gentler" qualities if one does not have to?

REFERENCES

Almond, G. A. 1956. "Comparative Political Systems," Journal of Politics IS, 391-409.

Althusius, J. 1603 [1995]. Politica: An Abridgcd Translation of Politics Methodically Set
Forth and Illustrated with Sacred and Profane E.xamples, trans, by F. S. Carney.
Indianapolis: Liberty Fund.

Barry, B. 1975. "Political Accommodation and Consociational Democracy." Britisk Journal of
Political Science 5, 477-505.

Cowcll, A. 1994. "In Italy, All or Nothing," New York Times, April 5.

Friedrich, C. J. 1968. Trends of Federalism in Theorx and Practice. New York: Praeger.

LaPalombara. J. 1987. Democracy, Italian Styk. New Haven: Yale Univcrsity Press.

Lijphart, A. 1969. "Consociational Democracy," World Politics 21. 207-225.

Lijphart, A. 1977. Democracy in Plural Societies: A Comparative Exploration. New Haven:
Yale University Press.

Lijphart, A. 1984. Democracies: Patterns of Majoritarian and Consensus Government in
Twenty-One Countries. New Haven: Yale University Press.

Lijphart, A. 1985. Power-Sharing in South Africa. Berkeley: Institute of International Studies,
University of California.

Lijphart, A. 1994. "Democracies: Forms, Performance, and Constitutional Engineering,"
European Journal of Political Research 25, 1-17.

Wheare, K. C. 1946. Federal Government. Oxford: Oxford University Press.