DEBATE POWER-SHARING, ETHNIC AGNOSTICISM, AND POLITICAL PRAGMATISM Arend Lijphart In his article 'South Africa: a consociational path or peace?', Rupert Taylor (1992) gives a firmly negative answer to the question he poses in his title. He rejects my recommendation of power-sharing (consociationalism) as the optimal method to achieve peace and democracy in South Africa on two grounds. First, he argues that my proposals are based on the assumption that South Africa's ethnic divisions are primordial in nature, that this assumption is incorrect, and that, consequently, power-sharing is not a realistic solution. Second, he criticizes my scientific methods and concludes that these methodological weaknesses constitute an additional reason for rejecting power-sharing. My response is also twofold. First, I concede that Taylor's critique is partly correct with regard to both primordialism and methodology. But, secondly, I shall argue that neither primordialism nor my methodological stance are necessary conditions for the possibility of instituting power-sharing in divided societies. Power-sharing does not depend on any prior assumptions concerning the nature of societal divisions, whether they are ethnic or non-ethnic, nor, in the case of ethnic divisions, whether ethnicity is primordial or instrumental and situational; what is more, power-sharing does not even require any prior assumption on whether a particular society is deeply divided or not. Neither does the estab- lishment of power-sharing depend on the advice of social scientists and the empirical evidence they have been able to submit; instead, the historical record shows that power-sharing has been instituted by politicians as a pragmatic political compromise without significant scholarly inputs. In short, while Taylor's objections to primordialism and to my methodology are partly well- taken, they are irrelevant as far as the possibility and promise of power-sharing are concerned. Primordialism vs Instrumentalism vs Ethnic Agnosticism Taylor is right that, in my earlier writings, I tended to accept the primordialist interpretation of ethnic divisions and ethnic conflict. More recently, however, I have been increasingly impressed with the empirical evidence of the variability and fluidity of ethnic loyalties in many instances. I was actually surprised to read 94 TRANSFORMATION 21 (1993) LIJPHART DEBATE Taylor's (1992:5) quotation of my quite recent description of ethnic differences as an 'unalterable fact' (Lijphart, 1989:14); that was a rhetorical hyperbole which I should have qualified more carefully. On the other hand, the difference between the primordial assumption that ethnicity is a 'given' and the instrumental-situational assumption that it is 'made' or 'taken' by politicians should not be overstated. That ethnicity as such is unimportant until it is used and given shape by political entrepreneurs is obvious- ly true, even trite. This means that, in theory, democratic politicians have the option not to appeal to ethnic or other major societal differences. But the temptation to do so is very strong, and politicians who unselfishly and high-min- dedly ignore ethnicity are likely to do so at their own peril, because they are likely to be defeated by more opportunistic colleagues. However, because of the fluidity of ethnic allegiances, it is better for power- sharing systems not to pre-determine the ethnic or other major components of a society. There is nothing in consociational theory that says that ethnic groups need to be described in formal and rigid terms. There are examples of power- sharing systems based on explicit, formal, and rigid classifications of ethnic groups, such as the 1943 National Pact in Lebanon and the 1960 Cypriot Constitution, but there are also examples of power-sharing based on informal and flexible coalitions and compromises of ethnic parlies, such as the Alliance and now the National Front in Malaysia. The Malaysian system has obviously worked much better than the Lebanese and Cypriot systems, and I have come to the general conclusion that it is preferable to allow a flexible self-determination instead of a rigid pre-determination of ethnic and other groups. Furthermore, I have come to believe that the optimal institutions for the self-determination of such groups are (1) elections by means of proportional representation and (2) the right of voluntary associations to organize their own, autonomous but state-funded, schools and cultural organizations. Such methods represent what may be called the agnostic approach to ethnicity: they allow us to 'agree to disagree' about which groups should be identified as the essential components of power-sharing systems in South Africa and elsewhere and even about what DonaldL Horowitz (1991:1-41) has called the 'metaconflict' in South Africa - the deep disagreement concerning what the true nature of the conflict is and whether South Africa can even be described as a divided society or not. Proportional representation permits the emergence and political representation of any group, ethnic or non-ethnic, and is not biased for or against ethnic parties. Another advantage of proportional representation is that it promotes the flexible adjustment of power and representation in response to any changes in the boundaries and strength of ethnic groups. And, finally, it has the advantage that it impartially allows the representation of 'new' minorities such as environmen- TRANSFORMATION 21 (1993) 95 DEBATE LIJPHART tal-protection parties, parties that advocate participatory democracy, and other groups that develop radically new perspectives on and approaches to politics. Having said all of this, I do not want to minimize the disagreement between Taylor and me concerning the importance of ethnicity in South Africa. Especially in the light of the pervasiveness and salience of ethnic differences and ethnic conflicts almost everywhere else in the world, it seems to me that it is unrealistic to assume that South Africa will be the 'grand exception'. Compared with most other multi-ethnic countries, South African ethnic differences do appear to be more muted in the 1980s and early-1990s. But there is a rather obvious explana- tion for this deviant phenomenon: the fact that the white minority government's widely despised policies of emphasizing racial and ethnic differences have been counterproductive. If this explanation is correct, we should expect a resurgence of ethnic differences when democracy has replaced apartheid. There is no way to prove whether Taylor is right or I am right at this point in time. My suggestion is that we wait ten to fifteen years; if full democracy will indeed be instituted in South Africa, if it is the kind of democracy that impartially allows the emergence of ethnic and non-ethnic parties, and if it persists until about the year 2005, my prediction is that ethnicity will then be a stronger and more pervasive force in politics than it is today. In the meantime, I suggest that we can be agnostic about ethnicity, but that we can agree about the need to institute democracy and, I hope, about the need to institute the kind of democracy in which both ethnic and non-ethnic groups and parties can flourish. Social Science and its Practical Relevance As far as my methodology is concerned, Taylor attacks me from two angles. On the one hand, he charges that I am not scientific enough and, on the other hand, that I am too scientific in the sense of being an 'empiricist*. I readily concede that many of my measurements are very rough. For instance, I frequently use the degree of pluralism in a plural society (or the degree to which a society is plural) as a key variable without having developed a fully operationalized, cardinal, measure of this variable.1 The reason, however, is not that this is something that I have simply neglected to do, but that this is a very difficult thing to do - something that, to my knowledge, no other social scientist has succeeded in developing. Faced with this obstacle, my choice has been to work with a rough nominal or ordinal measurement (such as plural vs non-plural societies, or plural, semi-plural, and non-plural societies) and to take the next step of investigating therelationshipof this variable with various other variables. I am convinced that science can make little progress if we get stuck at the level of measurement problems. Moreover, if we want science to have social relevance, we should not listen to the scientific purists, but instead, when narrow measurement and 96 TRANSFORMATION 21 (1993) LIJPHART DEBATE classification problems tend to get us bogged down, to bypass these problems and to move on to the bigger questions. With regard to the second criticism, I concede that my approach to social science entails the search for patterns and regularities. Inherent in this approach is not just an open mind but an unremitting duty to investigate 'deeper meanings' in the sense of alternative or additional variables that may be able to explain social phenomena better. It is in this spirit that I formulated my nine favorable conditions for power-sharing; these can help in making explanations and predic- tions about the establishment of consociational democracy and of its success more accurate. (To criticize these because they are not perfect predictors, in the sense of necessary and sufficient conditions, is to fall prey to the self-defeating scientific purism mentioned in the previous paragraph.) I also believe that there are useful practical lessons to be learned from both the general pattern and specific instances of power-sharing. But whether these have much actual influence is very doubtful. In the case of my own power-sharing advice for South Africa, it has even had a negative influence as a result of the National Party's misuse and abuse of the concept of consociation as an argument in support of the new 'tricameral' constitution in the early 1980s. On balance, I think that my ideas have played a slightly positive role in the democratic development of South Africa - but this assessment may be biased by both wishful thinking and immodesty. The more important point that I want to make is that the historical record does not show that either social science generalizations or specific empirical examples have played a significant role in the establishment of power-sharing systems. Let us take a look at the ten major examples of the institution of power-sharing: in the United Province of Canada (1840), the Netherlands (1917), Lebanon (1943), Switzerland (1943), Austria (1945), Malaysia (1955), Colombia (1958), Cyprus (1960), Belgium (1970), and Czecho-Slovakia (1989). What is striking about these ten quite different cases - spanning four continents and almost a century and a half - is that power-sharing was developed repeatedly and independently as a method of resolving and preventing deep conflicts. This shows that power- sharing has a compelling logic. But this logic was not helped in any way by social science or historical knowledge of previous examples of power-sharing. For instance, there is no evidence that the Malaysians knew anything about the Lebanese precedent when they developed their power-sharing system, or that the Colombians were aware of the Dutch model when they designed theirs. South Africa appears to be a deviant case because South Africans have made great efforts to examine the potential relevance of social science theories, including consociationalism, and of foreign examples. However, my interpreta- tion of the current South African political scene is that South Africa is moving TRANSFORMATION 21 (1993) 97 DEBATE LIJPHART in the direction of power-sharing democracy, not as a result of all these admirable efforts, but mainly as a result of the dynamics of negotiating a solution in a highly conflictual society. In other words, I believe that, as in the other ten cases, power-sharing is once again being re-invented in South Africa. What consocia- tional theorists have been able to discover appears to have little relevance. Political pragmatism and the inherent logic of power-sharing are the key ele- ments. Conclusion Because power-sharing is not dependent on any particular notions of ethnicity and social science, it would be possible for Taylor and me to disagree on these two matters and nevertheless to agree on the desirability of power-sharing. However, my guess is that Taylor has additional objections to power-sharing, two of which he refers to in passing. I should like to comment briefly on both of these. First, he expresses a strong preference to participatory democracy, which appears to be in conflict with the 'elitist' nature of consociational democracy. I now believe that, especially in my bookDemocracy in Plural Societies (1977:49- 50, 53-54), I overstated the role of strong leadership and mass deference in consociational democracies. I now think that the two principal types of repre- sentative democracy - consociational and competitive or majoritarian (British- style) democracy - barely differ with regard to the political power of elites and that the major difference between them is the elite's governing style. This also means that participatory democracy is equally compatible with the two major types. The one word of warning that I should like to express is that political participation spells political power (which is, of course, one of the reasons why participatory democrats favor greater participation), but that social science research has also found that political participation is highly correlated with social status and income. Therefore, especially in a country with extreme socio- economic differences like South Africa, one must be careful that participatory democracy does not lead to even more privileges for the already privileged strata of society. Secondly, Taylor (1992:5) writes that consociationalism 'denies a common humanity'. This is probably a rhetorical hyperbole on his part, but, to set the record straight, my own characterization of consociationalism is that it recog- nizes both a common humanity and important differences between human beings. As a normative theory, consociationalism recommends methods that help to manage these differences in such a way that the common humanity is given optimal scope. 98 TRANSFORMATION 21 (1993) LIJPHART DEBATE NOTE 1. The graph that Taylor (1992:8) reproduces (Graph 1: Probabilities of Success of Two Normative Models) from my boot Democracy in Plural Societies (1977:237) was not meant to 'prove' anything, but simply aimed to illustrate the logical point that, if we assume that the likelihood of success of consociational and adversarial methods decline and eventually reach the zero-point as pluralism increases but that consociational methods work better, even if only slightly, there is an area where consodationalism becomes the only possible, albeit far from a failure-proof, method. REFERENCES Horowitz, DL (1991) - A Democratic South Africa? Constitutional engineering in a divided society (Berkeley: University of California Press). Lijphart, A (1977) - Democracy in Plural Societies: a comparative exploration (New Haven: Yale University Press). Taylor, R (1992) - 'South Africa: a consociational path to peace?', in Transformation, 17. TRANSFORMATION 21 (1993) 99