previous chapter
Chapter Seven— The Transition to Democracy and the Problem of Hegemony
next section

Chapter Seven—
The Transition to Democracy and the Problem of Hegemony

Since 1967, power has never passed from one elected government to another in Africa. Before 1967, although several states held free elections won by incumbents, almost no civilian government ever handed over power to a duly elected government from among the former opposition. There were, to be sure, two short-lived cases of peaceful alternation. In Benin, from 1957 to 1965, several civilian coalitions alternated in office, although usually with one of the coalition partners constant from one government to another, until a military coup ended the process. In the Sudan, power passed peacefully in 1956 from one party to a coalition of two others. In the elections that followed in 1958, a new coalition—including one party that had served in the former government—formed a majority. A coup later in the year put an end to that government and to alternation in office. There was also an aborted attempt at alternation in Sierra Leone in 1967. A new government had barely been sworn in when it was overthrown by the military. After that, alternation ended. No African political party has ever again moved from the opposition to the government benches after an election.

Rather than turn power over, elected governments transformed themselves into single-party regimes. In several countries, continuing opposition electoral strength precipitated the declaration of a one-party state. As we have seen, in those few countries where an opposition functions and free elections are held—Senegal, Botswana, and the Gambia—the structure of cleavages is such as to give the ruling party a dom-


inant position, free of any real prospect of losing power at the polls. If democracy includes the possibility of participation or alternation in office, as it does in most conceptions—including Nigerian conceptions[1] —the African background against which South Africans write is not propitious.

As a matter of fact, three of the five long-standing Asian democracies, Japan, Malaysia, and India, have also been one-party-dominant systems through most of the post-colonial period. The other two, Sri Lanka and the Philippines, have had more difficulty maintaining stable


institutions. Multiparty competition with alternation has been rare, albeit not impossible.

These are chastening facts. Of course, the future of South Africa will be determined mainly in South Africa, not elsewhere in Africa and certainly not on other continents. However, some of the reasons for the failure of democracy elsewhere in Africa may also be present in South Africa.

Possibilities Versus Probabilities

In the preceding chapters, we have oscillated between asking whether democracy is possible in South African conditions and whether it is probable. It is now necessary to examine more closely the gap between what is possible and what is probable and to consider whether anything can be or has been done to enhance the probability that democracy will emerge.

In Chapters 1 through 3, I enumerated some of the difficulties that confront democratic planners in South Africa. These include a profound dissensus about how South African society should be understood and transformed; a metaconflict that, at the very least, constrains political innovation; contenders for power who entertain hegemonic, rather than power-sharing, aspirations; ethnic as well as racial differences that will be politically significant; an array of suspicions of other players, misconceptions about the nature of majority rule, and pretensions to an exclusive place in the state; the need for change to take place rapidly; and the absence of structural features like a strong, autonomous civil society with a stake in maintaining democracy.

In Chapter 4, we reviewed a variety of proposals for a democratic South Africa, concluding that some are not apt for a racially and ethnically divided country, that some are not apt for the particular ideological divisions that sit atop the ethnic and racial divisions, and that some are not apt for either. In the process, we saw that the ideological dissensus limits South Africa to a fairly narrow band of political institutions that would have a claim on legitimacy. We also examined a number of common mistakes in negotiating arrangements for divided societies. At least some of these errors are likely to be repeated in South Africa.

The burden of much, although not all, of the material in the first four chapters suggests that inclusive democracy is improbable in South Af-


rica. In that respect, South Africa is by no means out of line. Not only is democracy unusual in Africa, but it is also rare in ethnically and racially divided societies more generally.[2] Such societies need special precautions if they are not to be overtaken by authoritarianism.

On the other hand, Chapters 5 and 6 make it reasonably clear that democracy is possible in South Africa. Democratic institutions can be crafted that are appropriate for the divided society South Africa will be. Such institutions make no compromise on the question of majority rule and provide room for democratic pluralism to flourish. The institutions are nonracial and nonethnic. Without racially or ethnically based electoral rolls, reserved seats, group vetoes, or homeland states, the institutions meet the requirements that flow from South Africa's ideological dissensus. They could have a claim on rather wide legitimacy, which they certainly would need if they were to endure, rather than merely being accepted as interim solutions, to be scrapped at the first opportunity.

The electoral systems that were recommended are premised on the crucial importance of a single question: Are there sufficient incentives to intergroup accommodation? This puts the accent on how institutions actually work, rather than on acquiring a particular kit or inventory of institutions. It substitutes mechanisms supportive of accommodation for prescribed devices to be used in all contexts. Most important of all, it puts the voters behind accommodation in a way that an emphasis on mere coalition formation—that is, seat pooling without attention to vote pooling—can never do. Putting the voters back into the center of the calculations insures that accommodation does not take place over their heads or behind their backs, that it is not just another deal. Accommodation only takes place with the voters' consent. If, in addition to measures to limit conflict, South Africa needs anything at all, what it needs most is consent.

That consent is by no means assured in South Africa. Among the other systemic features a democratic South Africa will enjoy is a fragmented, potentially polarized party system. Institutional design needs to take account of this feature. Like electoral rules that encourage vote pooling, a central political officer, a president, elected on such a formula, would reinforce the centripetal pulls of the system. And just as


the electoral system changes the incentives, federalism would change the context of politics, so that state arenas, state loyalties, and substate competition would be activated. Unlike a unitary system, with a single arena in which intolerance can be writ large—which is roughly the situation from which South Africa is emerging—a federal system would permit a variety of local adjustments and might impede a single pattern of domination from emerging. The institutions that have served White interests, or narrower Afrikaner interests, are, in short, not appropriate to the heterogeneous polity South Africa must and will become if it is to escape a new domination. Fortunately, a new set of utterly group-neutral institutions is available.

Attractive as these institutions are, there is no blinking the difficulties attending their adoption and implementation. Vote-pooling electoral systems and interparty competition for votes are not easy to envision in the many townships where, in the 1980s, one or another organization—generally the ANC, AZAPO, or Inkatha—claimed exclusive control and ran its opponents out.[3] Perhaps peaceful competition is not altogether impossible. There have been some spontaneous street committee elections, in effect revolts against the bullying rule of young comrades. But people have become accustomed to local tyrannies.[4]

Or consider the persuasion it will take to convince the extraparliamentary opposition that federalism is not just a cleaned-up version of the homelands policy. The bias against federalism is so strong that, according to a sophisticated UDF leader who concedes its possibilities, it is not worth the immense effort it would take to change minds.[5] Many such innovations, it is thought by some UDF leaders, could be adopted at a later stage.[6] However, where a great disjunction exists between one regime and the next, "attitudes toward the legitimacy and efficacy of the new regime are quite likely to be permanently shaped by the initial steps."[7] Moreover, once new institutions are in place, political actors


adapt to them. After political parties are established, for example, it will be difficult to alter their position by attempting to change the electoral rules. "The party strategists will generally have a decisive influence on electoral legislation and opt for systems of aggregation most likely to consolidate their position. . . ."[8] The same, of course, goes for federalism, presidentialism, and other measures reallocating powers. Seymour Martin Lipset and Stein Rokkan speak of "the freezing of political alternatives"[9] in the formative phase of new regimes. In short, since interests will crystallize quickly around the first set of institutional innovations, it is a chimera to think that much flexibility will be retained thereafter.

If there is one proposition that commands wide assent among theorists of democracy, it is that democracy is no single form of government but "an array of possibilities," a "system of processing and terminating intergroup conflicts" without foreordained outcomes, a way of "institutionalizing uncertainty," an arrangement in which many people can pursue "widely varying goals over time."[10] This is, of course, one reason why rigidly predetermined ascriptive majorities and minorities can hardly be said to be conducive to democratic rule. The indeterminacy prescribed by these conceptions of democracy implies that no group should indefinitely be denied the opportunity to participate in government.[11] But to assure that opportunity, to make the system fluid, requires a configuration of institutions that South Africans have not yet embraced across the many cleavage lines that divide them.

Reviewing all this evidence, then, the conclusion is inescapable that democracy is possible but improbable in South Africa. The institutions for democracy are available, but they are not likely to be adopted or, if adopted, made to work. Can these conditions be altered? Is it feasible


to convert possibility into probability or at least to reduce the gap between the two?

Clearly, some conditions are not alterable in the short run. The need for rapid change, for example, is a given. A weak civil society cannot be strengthened overnight. Fundamental cleavages cannot be abolished, and they certainly should not be wished away. Yet, proponents of democracy need not be content with all aspects of the environment they confront. It is plainly inadequate for constitutional engineers to produce an appropriate design and let it languish in inhospitable conditions, some of which might yield to deliberate effort.

An analogy to inhospitable international conditions is suggestive. In the struggle to create international cooperation in the face of rewards for noncooperation, international actors have worked to change the structure of the situation—to provide incentives for cooperation and penalties for noncooperation, to engage in reciprocally rewarding behavior, to create international regimes with more stable rules and procedures, and to gain acceptance for new, cooperative norms.[12] What we have for South Africa is a set of new norms thus far awaiting actors who are willing to put them into practice by committing themselves to appropriate political behavior. Can anything be done, comparable to what has been done internationally, to enhance the prospects for cooperation and democracy, rather than conflict and authoritarianism, in South Africa?

The crucial conditions in need of alteration concern the aspirations of the main political actors and the prospects for the longevity of apt new institutions. Although these two are related, the first brings us back to the difficult problem of hegemony raised in Chapter 1, and the second focuses particularly on the process by which agreement is reached.

General Models of Democratic Commitment

The willingness to share power in a democratic regime cannot be taken for granted. Some actors may see the choice as simply to dominate or be dominated. Some may not value democracy, including the prospect of alternation in office. They may think that they have earned the right to rule, or they may believe democracy is too cumbersome a framework


in which to achieve transformative objectives. In any of these cases, they may aim at hegemony rather than democracy.

In South Africa, it has been suggested that one "contending regime model" is essentially authoritarian,[13] and so the hegemony problem exists. (A problem also exists, in a different way, with respect to the White right wing, about which I shall say a little more later.) The problem is, of course, exacerbated by the prevalence of undemocratic, single-party or military regimes in most southern African countries, with the exception thus far of Botswana, Swaziland (a monarchy), and Namibia (a dominant-party system but not yet a single-party system). For all the contending organizations in South Africa to forswear hegemonic aspirations is for them to renounce what has become nearly normal in most of the region and, for some of them, as we saw in Chapter 1, also to renounce a strand in their ideological commitment to socialist dictatorship. For the African National Congress, the matter is compounded by the overlapping composition of much of its leadership and that of the South African Communist Party (SACP), a party that has not been enthusiastic about the transformations toward liberal democracy in the Soviet Union and Eastern Europe.[14] And, for all of the extraparliamentary opposition, there is a further problem. If there has been a "liberation struggle" in South Africa, it may seem at best an ironic result to have to share power with those from whom liberation has been won. Not surprisingly, revolutionary turnovers are not likely to lead to democratic outcomes.[15]

The prospect of hegemony is, then, to be taken seriously, as nervous calls for multiparty democracy suggest.[16] Perhaps the first thing that needs to be said about this is that single-party systems are not democratic. Such a view is held even by those few who think that democracy is a system that can grow up only in certain limited historical periods,


such as early capitalism.[17] The obvious reason for the undemocratic character of one-party states is that contestation, the prospect of alternation, and even the possibility of exchanging opinions in an autonomous organization are all denied.[18] Many African single-party rulers have stayed in office for decades. The less obvious reason is that, in ethnic terms, such regimes tend not to be inclusive and are sometimes narrowly exclusive, because they no longer need to build electoral majorities.[19] For South Africa, there are straws in the wind: the strong affinities of some contenders with other one-party regimes, the climate of fear that has overtaken many townships, the frequent critiques of liberal democracy, and even the defense of summary executions of collaborators against criticism from human rights organizations.

An examination of the conditions that prompt the creation of one-party regimes in Africa suggests that those conditions may also be present in South Africa. In much of Africa, the single-party state emerged in response to party competition along ethnic lines. A ruling multiethnic party was threatened with erosion of its clientele. Ethnically based parties grew on one or both flanks of the ruling multiethnic party as sentiment polarized. Fearful of both the polarization and the shrunken base, the ruling party then declared a one-party state, outlawed the competitor parties, and thereafter purported to represent the interests of their former adherents.

Needless to say, such moves do not abolish ethnic conflict—they only mask and silence it.[20] Among other reasons, the former opposition parties and the ethnic groups they represent are rarely well integrated into the single party that ostensibly welcomes them. Established party functionaries make sure of that. What Fred M. Hayward and Jimmy D. Kandeh conclude for one-partyism in Sierra Leone is apt more gener-


ally: "The multiparty system, as a mechanism for electoral competition and change, has been viewed in many cases in Africa . . . as responsible for the intensification of ethnic and communal conflicts. As this [Sierra Leone] case demonstrates, the creation of a one-party state does not necessarily diminish such conflicts and may in fact intensify them by eliminating mechanisms designed to prevent such conflicts and the abuses leading to them."[21] Under the cloak of one-partyism, ethnic groups and cliques within groups are able to attain hegemonic influence that would probably be checked in openly democratic systems, where narrowing of a base of support would incur an electoral penalty.

To be sure, at least the Charterist contenders in South Africa, particularly the African National Congress, are multiracial and multiethnic. But so were many anti-colonial movements. After independence, some groups, often those less well represented in the umbrella movement, fell away, forming their own flank parties and eventually precipitating the end of party competition. South Africa already has several competing parties, with varying ethnic bases, and is as plausible a candidate as any other for a one-party, hegemonic regime.

How, then, should those who are interested in a democratic South Africa think about restructuring the situation to render hegemony less likely? With few exceptions, none of the major bodies of literature on transitions to democracy confronts the problem of hegemony from inside the regime. Retrospective explanations are available for why and how regimes become democratic.[22] Theories have also been devised for returning to democracy by inducing the military to withdraw from power and then keeping it out during the transition to democracy. Others are concerned with the ability of regimes to prevent oppositions from turning against the democratic system as well as to prevent anti-system opposition parties from taking over the system from the outside.[23] Backsliding to authoritarian rule has not been ignored, but much of the prodigious work on democratization in Latin America and southern Europe focuses principally on keeping the armed forces out and secondarily on keeping the opposition loyal.

On few other issues of comparative politics have interregional differences in focus been as much in evidence. Despite the prevalence of au-


thoritarian tendencies within new regimes and despite the African single-party state, hegemony from the inside is the least systematically treated issue of the transition to democracy.[24] The principal obstacle to democracy in Africa is thus hardly noticed in the transitions literature. The commitment of a new regime to democracy is the very thing that often cannot be taken for granted.

Ideological, racial, or ethnic polarization, on the South African model, is an unfavorable condition for democracy, but polarization can also lead to democratic accommodation. In Dankwart A. Rustow's model of transitions to democracy,[25] the decisive step to a democratic regime is preceded by a long period of struggle and polarization. Then, either at one decision point or several, leaders of the contending forces agree to accept what divides them and live with it in a democratic framework. In Sweden, this decision was facilitated by a great compromise. Universal suffrage, sought by the challengers, was accepted by those already within the system, in exchange for an electoral system that confined the former outsiders to a proportionate share of seats.

As Rustow and other writers suggest,[26] a democratic accommodation entails the creation of institutions to prevent vulnerable social forces from losing out to a new authoritarian regime. But what decides the participants on a commitment to the new democratic system? What cements the accommodation and overcomes hegemonic aspirations? Here we are truly in terra incognita. The possible models are many, and each suggests a different strategy.

One possibility is a deterrence model.[27] If power is located in several places, actors may hesitate to use their share of power in ways that are contrary to their commitments. In Portugal, for example, the armed forces, which held independent power during the transition to democracy, from 1976 to 1982, served as a guarantor of the transition.[28] A variant of this can be put in cost-benefit terms. Democracy may be permitted to the extent that a regime believes the attempt to coerce the


opposition will either fail or be too costly to justify the attempt.[29] To prevent retrogression to authoritarian rule, therefore, one strategy might be to make the success of any such attempts problematic and to make even small breaches of democratic norms transparently obvious.

A second way of thinking about the matter involves habituation . If democratic institutions can be made to work to the common advantage of the major actors across the divide between government and opposition for a long enough period for democratic habits to take hold, those habits will render backsliding less tempting and, should it occur, more aberrational and less legitimate than it would otherwise seem.[30]

A third model is premised on reciprocity . The literature on transitions to democracy in Latin America and southern Europe is suffused with discussions of the utility of "pacts," agreements among crucial actors that commit them to democracy. The pacts contain guarantees of mutual security and tangible rewards for cooperation in the process—in some formulations, rents to participating organizations, which are privileged over their rivals, so that democracy is underpinned by a form of corporatism.[31]

A fourth view emphasizes changing belief through learning. Hegemony is a problem, because not all the actors believe in democracy. They need to learn that the consequence of a failure of democracy (and power sharing) in a divided society is a long period of destructive conflict. Learning therefore focuses on the advantages of cooperation and the costs of trying to secure hegemony.[32] In contrast to habituation, which also involves learning, the changes in belief envisioned by this view take place at the inception of the new democratic regime.

All of these models are ideal types, and there is overlap among them. It is well known in learning theory that the acquisition of knowledge proceeds more readily when learning is rewarding. The internalization of democratic norms may thus be reinforced when deterrence makes


resort to undemocratic behavior dangerous and reciprocity provides a payoff for conformity to democratic norms. In spite of the overlap among the models, however, none of these models—not even all of them together—can provide a sure path or even a probable path to democracy in South Africa. All of them set up fences higher than can be scaled, quite possibly higher than need to be scaled. Later in this chapter, I shall suggest a less demanding set of possibilities.

From Hegemony to Democracy:
Blocked Paths

The one model that can be pushed to one side for the moment is habituation. Democratic habits take hold as democratic practice proceeds. Habituation is therefore cement that binds democracy at a later stage, perhaps even in a later generation than in the formative generation.[33] For the formative period, the question is whether actors will commit themselves to the "act of alienation of control over outcomes of conflicts that constitutes the decisive step toward democracy."[34] This is a difficult step to take, because all the forces seeking to bring down an undemocratic regime want simultaneously to create conditions that will favor their specific interests in the regime that replaces it.[35] But democracy embodies a large measure of indeterminacy. Habit, then, cannot solve this problem, because pre-democratic habits are, by definition, not democratic and because the impulses of so many actors in the process are to restrict the uncertainty that democracy brings.

Of course, deterrence, reciprocity, and learning do have something to contribute to securing commitments to democracy. The existence of powerful forces with the ability to retaliate if democratic norms are breached might induce a degree of conformity to democratic norms. The commitment of other actors to forswear the use of undemocratic methods might elicit a reciprocal promise to do likewise. Learning the hazards of conflict and the possibilities of cooperation might reduce the temptation to high-risk, hegemonic strategies. It is, therefore, worth taking a closer look at each of these approaches.


Deterrence can be produced by the structure of forces already in existence or by deliberate design. In South Africa, the Congress of South African Trade Unions (COSATU), a strong union that has enjoyed close relations with the UDF and the ANC, might be wary of an attempt to establish a one-party state. Such regimes have generally not been receptive to expensive wage demands emanating from trade unions or even to the principle of free trade unionism. The disciplined power of COSATU, whose members have shown their cohesion in strikes and protests, could act as a deterrent. In this case, a force with some checking power is thus already in existence.

Not many such forces can be identified, however. I have suggested in Chapter 6 that a decision in favor of federalism would produce more such forces and raise the costs of a potential move to authoritarianism. This would be deterrence by design. The problem is that the argument is circular. Knowing that federalism creates a more complex system that makes authoritarianism problematic implies that those not yet committed fully to liberal democracy will resist creating a federal state or will attempt to convert it into a sham federalism, precisely to reduce its inhibiting power.

And so we are back where we began. South Africa does have some potential counterforces. Were they to be given scope to operate, there might be enough pluralism to sustain democratic complexity. But potential hegemons will oppose providing the requisite scope right from the beginning.

The same goes for reciprocity in the form of pacts, guarantees of mutual security, and institutional compromises. There is no doubt that, in some settings, agreements have been forged in circumstances that, without them, would have been uncongenial to moving forward to a consolidation of democracy. Sweden's great compromise, mentioned earlier, is an example. Another was consummated in Portugal in 1975, when revolutionary attempts to redistribute land provoked a right-wing reaction that threatened to reverse the democratic process. At that point, the Socialist-led government of Mario Soares crafted an agreement returning most of the seized property to its former owners and ending rural anarchy, but ratifying state ownership of already-nationalized industries. The effect was to provide a new sense of security to the threatened right and to consolidate democracy at the expense of redistribution. Not only did such an agreement work in Portugal, but, suggests Nancy Bermeo, in every case where a transition to democracy was successful, "dramatic redistributions of property were postponed, circum-


scribed, or rolled back."[36] The agreement, explicit or not, was to trade more mass democracy for less mass economic welfare.

Again, such conclusions may reflect the European and Latin American focus of most studies of transitions to democracy. In Europe and Latin America, the main threats to democracy, albeit not the only threats, have come from the right. The assumption has been that force is not controlled by the left in a transition situation, so that the left is, by default, committed to the democratic path to achieving its objectives.[37] That has largely been true in Europe and Latin America, but it is not uniformly true.

Explicit agreements to protect weaker elements in the polity and to prevent authoritarianism have been recommended in general, and they have been recommended for South Africa.[38] The United States also managed a regime change from a troubled confederation to a strong federal union in 1787 by means of, among other devices, some great compromises, especially those between small states and large states.

In the United States, however, and arguably elsewhere, there was an underlying consensus on the utility of liberal democracy, even though there was no consensus on the institutional forms it should take or the extent to which various interests should be protected. In South Africa, that underlying consensus is the very thing in issue. What Guillermo O'Donnell and Philippe C. Schmitter call "contingent consent."[39] or agreement to compete so that temporary winners do not use their power to impede those who win election in the future from taking office, is the object of a possible exchange that would indeed provide mutual security. But if one party to such a transaction thinks it can find equivalent or better security without making such an agreement, is that belief not an insuperable obstacle to consummating the transaction?

I have discussed the pitfalls of deals and contracts in Chapter 4, and I shall return to the subject in connection with the negotiation process. Lying in wait, years down the road, are many more hazards to the contractual mode of arranging political affairs than have been appreciated


by the proponents of pacts. Some of the same pitfalls have been identified in connection with reciprocity as a strategy for achieving international cooperation.[40] But, whatever their ultimate advantages and disadvantages in the democratization process, it seems clear that such exchanges cannot by themselves cause democracy to be seen as the best alternative.

Before we leave deterrence, reciprocity, and habituation, perhaps we should consider the possible efficacy of a combination of all three. Suppose agreements are struck that embody some of the characteristic payoffs that are said to accompany pacts and also that those pacts include strong, publicly stated commitments to liberal democracy. Suppose further that the accompanying constitutional arrangements embody powerful elements of deterrence, including a Madisonian system of checks and balances, interim guarantees of democracy (perhaps involving the armed forces, on the Portuguese model), and clear rules of the political game that would make departures from democratic norms embarrassingly and dangerously obvious. If we postulate adoption of these arrangements, along with appropriate institutions to mitigate racial and ethnic conflict, would we be warranted in speculating that habituation could then take over and propel the regime into a democratic future?

My own view is that even this configuration would not be enough. It is one thing to ask committed socialists to understand the advantages of the free market, particularly when the social costs of a market economy can be reduced with welfare policies. A remarkable number of seemingly committed socialists have been willing to make this conversion. But the equivalent conversion to the political market is more farfetched for committed authoritarians or even weakly committed democrats. For one thing, the indeterminacy of the political market can oust them from power altogether. Economic policy may produce the same result, but such a possibility is not an intrinsic part of the policy choice, as it is with democracy. For another, the benefits of democracy are simply not as palpable as are the benefits of the free market. Political freedom does not have quite the same feel to it as economic welfare does, and politicians are less certain that they can claim credit for liberty than they are that they can reap rewards from prosperity. This is particularly the case in a society where people value their own liberty but are ambivalent about the liberty of other social groups. And so the arrangements hypothesized here, entailing interlocking parts of deterrence and


reciprocity, with a hope of habituation, seem altogether too contrived, too confining, perhaps too open to insincere assent, to produce democratic habits.

If strong medicine is probably insufficient to overcome deficiencies of belief, perhaps we are back to learning. Can the dangers of a quest for hegemony be learned? Can the utility of contingent consent be impressed upon the participants? When and how?

Interpreting the findings of studies in southern Europe and Latin America, Bermeo emphasizes "value change,"[41] by which she means that the experience of authoritarianism often—though not always—inclined many voters and many politicians to moderation, to self-limiting compromise, when the time came for a transition to democracy. Political actors had learned that extreme demands and unsettled conditions might prompt renewed military intervention. Lessons like these may be essential to democratization, which can only be brought about with "heroic forms of restraint and cooperation," to use Robert R. Kaufman's language.[42]

Whatever opponents of authoritarian regimes may have learned in many Latin countries, the opponents of colonial authoritarianism in Africa obviously learned something quite different. The point is so transparent that it is rarely stated: most post-colonial African leaders have had only a shallow belief in democracy. Quite accurately, Naomi Chazan describes Ghana's brief experiment with democracy after the British departure as "half-hearted and short-lived. Ghanaians had had little practical experience with this model under colonial rule until the postwar period. . . . In Ghana, as elsewhere on the continent, a British style of parliamentary democracy retreated quickly along with its colonial originators because the new leaders had little commitment to uphold its precepts."[43]

To be sure, independent South Africa has had far more experience with parliamentary democracy, but it has been democracy for Whites only. There is little reason to think that the excluded majority has somehow been able to overcome the knowledge that parliamentary institutions have been the source of discriminatory legislation and instead managed to develop an affinity—a positive preference—for a now-


inclusive democracy of the very sort that has been so conducive to the experience that is to be transcended. As I have shown in Chapter 3, survey data indicate that a great many South Africans want a racially and ethnically inclusive polity, but the data provide no evidence of commitment to any particular arrangement of democratic institutions.

Neither is there any reason to think that, confronted by recalcitrant White authoritarians, the extraparliamentary opposition has internalized the lessons so well learned in Latin America—that restraint and cooperation, commitment to mutual guarantees, moderation and compromise, are preferable to confrontation with former oppressors. A finding that value change preceded successful transitions to democracy is actually a pessimistic omen for the many cases in which such a transition has not taken place and in which democratic values have not yet become preeminent. Learning may still be the appropriate model, but, for South Africa, it will not be the learning that took place during the period of exclusion. That learning does not seem particularly congenial to the position that democracy and power sharing constitute the preferred alternative. Instead, if learning is to play a role, the relevant value change is that which takes place during the democratization process itself. This need not be fatal. Perhaps, as Giuseppe Di Palma says, "genuine democrats need not precede democracy. . . ."[44]

One thing that could be learned during the planning for inclusive democracy is that the price of a regime change is a commitment to democratic institutions on the part of all contenders for power, since the White regime will not change without such a commitment. I have suggested earlier that this is not likely to be a fruitful way to proceed. Standing alone, such a condition will encourage, not a change of heart, but a change of tactics. South Africa needs more than paper guarantees.

On the other hand, as I mentioned in Chapter 1, the problem can be cast in terms of learning just how explosive and inimical to everyone's interests intergroup conflict and violence can be in a divided society. Advocates of consociational democracy have argued that the path to such an agreement begins with a recognition of mutual vulnerability, facilitated by "a glimpse into the abyss of violence. . . ."[45] As I have argued earlier, the best time for the appreciation of mutual vulnerability is after serious violence, when vulnerability has been demonstrated clearly. The Nigerians and Ugandans are witnesses to the point. Nevertheless,


the requisite learning about vulnerability, it is said, can occur during negotiations.[46] The key, then, is enlightened self-interest. But how is "enlightenment" to take place if an actor is "unenlightened," that is to say, unconvinced?

The negotiating table is an unlikely setting for learning. The time frame is too short for the requisite shifts to take place. Furthermore, what we know about cognition suggests that "people assimilate new information to existing beliefs; they are 'cognitive misers.' Until and unless the information is sufficiently discrepant, they are unlikely to 'learn.' Leaders are most likely to learn when they repeatedly encounter strongly discrepant information that challenges fundamental beliefs. Only then are they likely to reformulate their beliefs about others and reframe the problem."[47] Unlike deterrence and reciprocity as models for overcoming hegemonic aspirations, the learning model does not have elements of tautology about it. Rather, it is, on empirical grounds, excessively optimistic.

The process of negotiation is also likely to be inappropriate for an emphasis on danger. The whole course of negotiation is inevitably fixed on reducing differences, creating trust and good will, displaying statesmanship, matching concession with concession. As a negotiation proceeds, the psychology of the event is hardly conducive to an emphasis on doomsday thinking, except in the thin, rhetorical sense of holding up the specter of what will happen if negotiations fail, the better to consummate whatever transaction is on the table. A bright future together is a spur to negotiation, and that is where the emphasis will be. Mediators, whose overwhelming incentive is to facilitate an agreement, cannot be expected to emphasize the disagreeable prospects that could only complicate the process. Successful mediators typically take pride in their ability to push thorny issues aside and concentrate on issues that lend themselves to resolution. For all these reasons, negotiations are unlikely to fix the minds of the participants on the sort of conflictprone polity to which they belong and on the sort of hostility toward which they might be headed. Negotiation is simply not structured to serve the purpose of changing values.

Theorists of negotiation do distinguish two phases: the tentative phase


of "prenegotiation" and the phase of negotiation proper that follows.[48] In the latter phase, a plan is formulated. First, however, distrustful perceptions of the opponent as enemy must give way to perceptions of the opponent as someone potentially able to cooperate and worthy of at least some trust. At the same time, during prenegotiation, efforts are usually made to convince the other side that concessions will be reciprocated and to prepare one's own side to accept the need for such concessions. It is true that prenegotiation is also said to be the forum for assessing risks and learning about costs.[49] But nothing so ambitious as a fundamental change of objectives or commitments is encompassed in such a formulation. On the contrary, a central function of prenegotiation is to limit the agenda for negotiation to problems that are soluble—within the actors' present frameworks—and the success of prenegotiation is said to depend on the existence of conditions that amount to an assessment, even before prenegotiation begins, that an agreement is probable.[50]

The usual negotiation, then, has modest goals compared to the goal of making completely new commitments, based on a model of changed values or new learning. Here it is instructive to compare with the two stages of prenegotiation and negotiation Pierre du Toit and Jannie Gagiano's three-stage model for reconciling differences in a divided society like South Africa.[51] Working backward, the third stage, "substantive bargaining," and the second stage, "preliminary bargaining," are recognizable as negotiation and prenegotiation. The first stage, called "bargaining about bargaining," is distinctive. The necessity for it derives from the "basic question": "whether adversaries who hate, fear and resent one another are likely to grant each other civil liberties, and whether they are likely to use their own freedoms to protect the freedoms of what they assess to be their deadly rivals. If not, how can they be persuaded to do so?"[52]


During the first stage, du Toit postulates a process of persuasion that the zero-sum view of the conflict is wrong, that "competing groups can do very badly together, with none emerging victorious," or "all the competitors can end up doing very well together."[53] The first stage is, then, the setting for learning all about the costs of conflict and the benefits of democratic power sharing. During this stage, the myths entertained by parties to the conflict are demolished; leaders become motivated to avoid the dangers of the conflict, and they therefore become committed to cooperation; a middle ground, a "contract zone," is created.[54]

Since there is in South Africa no consensus "over the defining characteristics of the state,"[55] particularly whether it will be a liberal democracy, this persuasion phase entails a herculean task, not comparable to the task performed during what is conventionally referred to as prenegotiation. How persuasion will be accomplished is necessarily, therefore, uncertain. The first phase is a black box.

To open this box, one possibility, as we saw in Chapter 1, is to seek analogies in the creation of international regimes in conditions of anarchy.[56] In such matters, undoubtedly, there has been learning, and there has been cooperation where there might still have been conflict. Here, too, negotiation features prominently in putting the new knowledge to work for cooperative ends. The new knowledge is exactly the sort that is necessary for divided societies. It consists of shared cognitions about joint gains that can be achieved by collaborative behavior but not by conflict behavior. Yet all of the learning does not take place during the negotiation process. Some negotiations will not yield fruit precisely because the politicians do not uniformly share the new cognitions that are underpinned by expert knowledge. In general, only where they do can the empathy that sometimes is created during negotiations help produce a zone of joint gains.[57]

Even if the lessons of international regime creation were otherwise relevant, two main caveats should be borne in mind. First, regimes are


created sector by sector. Regulation of international financial transactions may be agreed to, even while a new regime for the oceans does not command assent. Second, the regimes are not only specialized but international. On both counts, they do not, for the most part, endanger fundamental survival interests. By contrast, a regime for the democratic governance of a racially, ethnically, and ideologically polarized society immediately implicates the core interests of its citizens. No legitimation in terms of expert knowledge will suffice.

In addition, racial and ethnic issues are often cast in terms of relative group advantage. Experimental evidence suggests that in such contests maximum intergroup differential is an outcome preferred to maximum joint profit, even where in-group profit in the joint profit situation exceeds in-group profit in the differential situation.[58] This is a payoff preference structure that is hardly conducive to cooperation under any circumstances[59] —which is why there is so much intergroup conflict in the world—and it is a structure that will require considerable insight into its risks for politicians to overcome the conflict behavior suggested by it.

One of the most promising strategies for enhancing prospects for international cooperation is lengthening "the shadow of the future."[60] The shadow of the future is the likelihood of continuing interaction among the players. It can strengthen cooperation, because conflict behavior can meet with retaliation when interaction extends into the future. A player with an interest in securing cooperation from another may attempt to lengthen the shadow of the future by decomposing payments (that is, paying on the installment plan) or by linking issues that are not yet linked.[61] This is, of course, another form of deterrence strategy.

The trouble with such a strategy in domestic politics is that sovereignty makes it difficult for a player without control of the state apparatus to withhold payment—difficult but perhaps not always impossible. More fundamentally, however, there is, in domestic politics, an analogue to the international strategy best calculated to limit retaliation and thus shorten the shadow of the future. In international relations, the strategy is preemptive war.[62] In domestic politics, it is the seizure of


hegemony, which brightens the future up considerably for the hegemon. In a sense, then, the reasons for cooperation to avert hegemony can be rendered irrelevant by seizing hegemony. This leads us right back to the question of why a potential hegemon might nonetheless wish to eschew hegemony.

In South Africa, the question can be made much more concrete by asking why a party espousing the claims of the Black majority should act in a conciliatory fashion toward any of the racial minorities—or toward ethnic or ideological minorities among Africans—by choosing institutions that would assure them of a democratic voice. Of course, the question becomes more pointed if we assume that one party, such as the African National Congress, could count on the support of a majority or near-majority of the South African population, enough at any rate to form a government. Hegemony then might not even require the use of undemocratic methods, such as the declaration of a one-party state; an undemocratic result might be achieved using majoritarian processes. The electoral systems described in Chapter 5 are premised on the need for incentives to moderation. But there is the prior question of what incentives exist to impel actors to choose and be faithful to such an approach. What, in short, are the incentives to adopt the incentives?[63]

To put the question in these terms has three advantages. First, unlike the belief model, the incentives formulation does not depend on the sort of transformation—from lacking a certain kind of knowledge to suddenly possessing it—that is not likely to take place in the time required to bring about a result conducive to cooperation. Self-interest, rather than cognitive change, becomes the focus of the inquiry. This is appropriate, since I have already argued that incentives and self-interest form the proper foundation for thinking about institutions that can simultaneously foster democracy and conflict reduction. Second, to cast the problem in terms of incentives avoids the rather artificial choice among competing models of deterrence, reciprocity, and belief. Various features of an actor's incentive structure can be the result of differing combinations of apprehensions, desires for rewards, and cognitions. Third, to focus on incentives is to recognize the contingent and potentially reversible character of the democratic venture. There is no point speaking of value change, for example, as if democratization involves crossing a threshold once and for all.


It is instructive to compare White and Black incentives. For Whites, there will be great reluctance to move to an inclusive democracy if there are risks of undemocratic outcomes or of economic decline. The fact that many Whites appear willing to take some risks is attributable to the delegitimation of exclusively White rule among many White elites and to the likelihood that the strength of the White position will otherwise decline as time passes.[64] Now contrast Black incentives. Black rule, which, I suggested in Chapter 3, can be conflated—erroneously—with majority rule, is not yet deemed illegitimate by all those who might benefit from it. And, if the White position is likely to decline over time, the Black position shows strong signs of improving over time.

The case for a Rawlsian social contract, for blind, original-position reasoning, depends upon considerable uncertainty about one's future position. The Nigerians, for example, were willing to make neutral arrangements to counter ethnic conflict, because their bitter experience had led them, not only to an extreme aversion to the dangers but also to considerable uncertainty about who would be its next victims. The same may yet prove true for the Ugandans. But the bitter experience of apartheid might seem to some Black South African leaders to point a different moral. As the White regime weakens and the Black position improves, the future dangers to Blacks seem less menacing. (Here I should underscore the word seem, for, as I argued in Chapter 2, the ultimate dangers for Blacks are very much greater than they seem as the White regime begins to relent.) The appeal of a democratic institutional framework that provides for power sharing and minority inclusion is accordingly reduced. On this view, then, White and Black incentives are asymmetrical.

The picture does not appear more promising if we review incentives that have moved leaders elsewhere toward democratic conflict reduction. Eric A. Nordlinger's list of four motives—external threat, desire to gain political power, fear of economic decline, and desire to avoid bloodshed—does as well as any.[65]

External threat is a condition commonly thought to move parties closer together. Certainly the impact of externally imposed economic


sanctions has had at least some force in South Africa. But it is most unrealistic to expect the prospect of adverse international reactions to prevent hegemony. After apartheid, the international salience of South Africa will decline precipitiously. International economic reactions, on the other hand, might well induce economic pragmatism. Some elements of the business community might then be persuaded to trade political hegemony for a market economy. This, of course, would reduce the incentives for potential hegemons to become democrats—the opposite of the desired effect.

The desire for power, which can only be acquired through compromise, is a possible incentive. However, that has not prevented other African leaders from altering their commitments in fairly short order by moving to single-party regimes.

The fear of economic damage from severe conflict and violence might be another motive for democratic accommodation, but political leaders in one divided society after another—among them, Cyprus, Lebanon, Sri Lanka—have been willing to risk extensive economic harm rather than take democratic precautions. And, of course, the same goes for the prospect of widespread bloodshed. Awareness of the danger is a large part of the "knowledge" I said earlier needed to be acquired. Unfortunately, that knowledge is generally acquired after the event.

No matter how we frame the question, South Africa appears to be another case in which there are more rewards for politicians to pursue both conflict and hegemony than to pursue accommodation and democracy. That does not mean that this will continue to be so under all conditions. What I shall argue for, before we are finished, is a possible model of changing incentives during the process of transition itself. This implies, first, that nothing so grand as a system of mutual deterrence or a major change in values need come into being for democracy to become more likely and, second, that the changes that take place are not preconditions to the process but integral parts of the ongoing process. At the outset, however, we need to see, concretely, how the incentives going into the process are not propitious. This, again, will underscore the long-shot character of the democratic gamble in South Africa.

Party Interests and Public Interests

If democracy is to emerge in doubtful or unfavorable conditions, it will have to emerge out of the new configuration of interests, positions, and strategies that take shape as events unfold. During this time, as we shall


see, every actor's incentives and commitments are shaped by those of every other, and so a few small changes can trigger larger changes throughout the spectrum. The process itself may unfold a new set of orientations toward the democratic regime.

To begin with, however, participation in the transition process may be grudging, and positions adopted may reflect the auspices under which the democratization process began. If the impetus comes from within the former regime, there will be, says Di Palma, "a 'left' that, even if sold on democracy (and not all will be), may not be sold yet on this democracy. It may fear the reappearance of the past in new guises. . . ."[66] This describes initial Black suspicion of what the South African government had in mind. On the other hand, Di Palma argues, a process begun from within the old regime has the potential for durability, since the initiators are in a position to offer reluctant sectors of the former regime an orderly transition. Moreover, initiators from within the regime may thwart the most dangerous, destabilizing reforms, provided they are able to retain the initiative and "set the agenda of democratic reconstruction."[67] All of this sounds as if it were written for—or by—F. W. de Klerk.

There is also a scenario in which the transition process is precipitated by radical mobilization and the radical left wants a "progressive" outcome, one not confined to liberal democracy but committed to a revolutionary change that obliterates the former class relations. This scenario, which, in some sense, also depicts the way the process began in South Africa, has the radicals claiming that other actors are "morally, politically, and economically bankrupt."[68] The political framework they want does not feature uncertain outcomes but rather radical socialist transformation. The democratic project is in jeopardy because, as Di Palma well puts it, "players who are reluctant to trust an open political game, preferring instead to protect themselves against their adversaries by girding democracy with their own invasive measures, may end up with a troubled democracy or worse."[69]

Between versions of these two alternatives lie South African prospects.[70] Much of the difficulty can be cast in terms of a wide divergence


between party interests and public interests, if by the latter we mean the requirements of a democratic order. The aim of winning may be so central as to preempt agreement to rules that do not guarantee winning but only guarantee competing. Commitment to democratic uncertainty will then be difficult to secure. South Africa has been no exception, for this is precisely the meaning of the government's long-standing willingness to share power without losing it. Commitments to democratic uncertainty are also difficult to obtain when ideology renders some of those with whom a player must compete illegitimate participants in the game. Finally, agreement to democratic rules that assure minority interests or influence is unlikely when a player representing majority interests forecasts that such rules may impinge on its prospects. Again, at the outset of the game, this depicts the South African situation.

The reluctance of the extraparliamentary contenders to appreciate and face the dangers of racial and ethnic conflict is underpinned by a strong, genuinely held set of ideological convictions. These beliefs label race and ethnicity artificial affiliations and stamp political action based on such affiliations as manifestations of false consciousness. Were this reluctance less pronounced, there might be some disposition to regard racial and ethnic inclusion as an important political value, and this view would then be conductive to the careful construction of inclusive democratic institutions. On the contrary, however, White political behavior based on racial affinity can easily be seen as an effort to retain privilege; and Black political behavior based on ethnicity—among, for example, Zulu in Natal—can be interpreted as the product of elite manipulation designed to perpetuate historically outmoded affiliations. Consequently, departures from hegemonic aspirations are made more difficult by the racially and ethnically differentiated character of several other likely contenders, against the background of ideological commitments hostile to organization based on ascriptive group.

These ideological predispositions are shared across Charterist-Africanist organizational lines. However, the organizations differ in their orientation to electoral politics. The Pan Africanist Congress and the Azanian People's Organization are both committed to some version of revolutionary liberation and uncommitted to electoral institutions. The ANC, on the other hand, is not at all hostile to electoral paths to power.


The positions of all of these organizations on this question are consistent with what is known about their share of popular support.

Of course, absent elections, it is difficult to gauge the support of organizations that might become political parties. First of all, survey respondents may express a preference for an organization that declines to participate in elections. Second, the process that eventuates in elections may change the support bases of the contenders. Third, variable electoral turnout, in rural versus urban areas, may be associated systematically with party preference. Fourth, the abstractly stated preference for one party or another may change when that preference must be translated into a choice among more and less attractive individual candidates. On all counts, South African surveys thus far have had a somewhat hypothetical character to them.

Nevertheless, virtually every survey shows stronger support for the ANC, its allies, and its leaders, than for any other organization. The ANC's support grew steadily through the 1980s. In 1981, 40 percent of urban Africans said the ANC would be their first choice in a parliamentary election. Inkatha would have been the first choice of 21 percent; the Black Consciousness organizations, of 11 percent; and the PAC, of 10 percent.[71] By 1985, another urban survey of Africans found that 49 percent of those asked who would make the best president for South Africa gave Nelson Mandela's name. Bishop Desmond Tutu had 24 percent, and Chief Mangosuthu Buthelezi had 6 percent. Again, the ANC-UDF forces emerged ahead of the others.[72] Comparable results were reached in another urban survey, in which Black support for Mandela, the ANC, the UDF, Tutu, and the Reverend Allan Boesak totaled more than 50 percent, compared to 8 percent for Buthelezi and Inkatha and 1–2 percent each for AZAPO and the PAC (including affiliates).[73] By early 1989, urban Africans gave Mandela 41 percent of their popularity preferences (interestingly, they gave P. W. Botha and F. W. de Klerk together 22 percent),[74] and careful observers attributed to the


ANC and its internal affiliates at that time a majority of African support.

The ANC's increasing support came clearly at the expense of Inkatha, the PAC, and the Black Consciousness organizations. Inkatha's fortunes outside Natal declined steadily as the ANC's improved. In 1977, Inkatha support was even with ANC support in the industrial Pretoria-Witwatersrand-Vereeniging (PWV) area. Each had about a third of Black support. By 1988, Inkatha's support in the PWV was below 5 percent, while the ANC's was 50 percent, even higher if affiliates are counted.[75] On the other hand, Inkatha remained strong in Natal, particularly in rural areas. In the late 1970s, after the Soweto killings of 1976, the fortunes of Black Consciousness organizations had risen dramatically. In suveys, students in Transvaal townships, including Soweto, chose Black Consciousness organizations and leaders over the ANC and Mandela by margins of 2 or 3 to 1.[76] After the UDF emerged dominant in the township protests of the mid-1980s, Black Consciousness was eclipsed. But the fortunes of all these organizations have fluctuated, and their leaders do not assume they are necessarily fixed.

All these surveys are of urban respondents only. Rural and homeland people, who constitute a majority of the African population, are harder to reach with sample surveys and have hardly been heard from in politics. In the few surveys in the 1980s with rural as well as urban coverage, the ANC remained ahead, but by a smaller margin; predictably, Buthelezi and Inkatha improved the showing they made in urban surveys, but to nothing like their former strength nationwide.[77] It bears emphasis also that the survey respondents were Africans. Coloureds and Indians, some of whom are found in all the extraparliamentary camps, are nevertheless, on the whole, decidedly more conservative. Many fewer Coloureds and Indians than Africans are likely to support the UDF or the ANC or, for that matter, any other Charterist or Africanist organization.[78] And, of course, the same is true to a greater extent for


Whites. After de Klerk announced the legalization of the ANC and the release of imprisoned ANC leaders in 1990, surveys of Africans outside rural Natal suggested that the ANC would have the support of more than half of all African voters in a universal suffrage election.[79] When rural Natal and non-Africans are added in, the ANC would have a plurality but not a majority of the vote.

Such estimates bear on the incentives the ANC might have to attempt hegemony or democracy. The ANC's preference for first-past-the-post is perfectly consistent with its position as the likely plurality choice of the South African electorate. Party interests may or may not converge with public interests. To take the clearest example, a party's position on electoral reform is a function of whether it will gain or lose from it. In Britain, where third parties habitually demand proportional representation, the two main parties speak of the plurality system in reverential terms. In Israel, where extreme proportional representation has had major adverse effects on the political system, the two largest parties may collaborate on a reform that benefits them and the political system while it harms small parties. It is infeasible to set aside party interests; and "exhortations to reach mutually desirable outcomes are often beside the point," for "undesirable yet 'rational' outcomes"[80] are distinctly possible.

A party with the plurality support the ANC might enjoy would inevitably benefit from a substantial seat bonus in a plurality electoral system. With 48 percent of the total vote in a first-past-the-post system, a party is virtually guaranteed a majority of seats, and a majority of seats is possible even on a 40 percent share of the vote.[81] With widespread electoral support and a fair number of three- or four-way contests, a party like the ANC could secure a very substantial majority of seats on a mere plurality of votes. These interests provide a strong foundation for the ANC's willingness at the outset to pursue an electoral route and also for its hostility to changes from first-past-the-post.

Similarly, the expectation of a large seat majority nationwide makes the ANC hostile to federalism. The regionally concentrated support of organizations like Inkatha means that federalism could only detract from the favorable prospects the ANC contemplates.


Here, then, there is a classical coincidence of interest and behavior. The extraparliamentary organization with the most support chooses a strategy of negotiation. The organizations that find themselves weak oppose negotiation, the results of which might entrench their weakness. As the ANC becomes more flexible, therefore, AZAPO and the PAC remain revolutionary, initially opposing negotiation with "the white minority settler government" (AZAPO) as a "sell-out" (PAC).[82] What each organization wants is consistent with its competitive position. As Jeremy Bentham said, "Interest smooths the road to faith."

Since I have already argued that a society like South Africa's should not have plurality elections and should have federalism, it is scarcely necessary to emphasize that the coincidence of party interest and behavior makes negotiation a perilous mode of institutional design. The same goes for the National Party's advocacy of group rights, which might also involve the equivalent of reserved White seats that the National Party, as the largest White party, could then capture. Group rights and reserved minority seats are, as I explained in Chapter 5, less valuable than is genuine influence in the political system. Once again, whatever the negotiators agree to is not necessarily what the society ought to have, because of the clear disjunction between party interests and public interests. Risk-averse actors prefer certainties that reinforce the favorable elements in their present position to the fluidity that a divided society needs and to the uncertainty of outcome that a democracy demands.

Party interests, however, are almost all we have to work with. Therefore, the issue is whether party interests can be harnessed to public goals—in this case, democratic goals. On this score, more interesting conclusions can be extracted from the ANC's interests and position.

First of all, a party with a strong preference for the seat bonus provided by plurality elections will probably find alternative voting to be its second choice. As I have noted, AV is usually seen as a modification of plurality elections, and many plurality parties are able to convert that plurality into a majority of votes through second and third preferences. Likewise, a party with widely distributed support may be receptive to electoral distribution requirements. These are hopeful signs for the ultimate arrangements that could emerge in South Africa.

Less hopeful, however, is another prospect associated with ANC in-


terests. The ANC's willingness to follow an electoral path may imply that what it contemplates is a secure ANC government, with a stable majority of seats, tolerating at the margins several powerless opposition parties: the National Party, perhaps even the Conservative Party, possibly Inkatha, the PAC, and AZAPO, if they should take the electoral path. The presence of active White parties could only enhance the ANC's base of support, by reactive mobilization. With periodic elections but no prospect of losing office, perhaps the ANC could comfortably run a dominant-party system on the model of Botswana, Senegal, and the Gambia—all of which survive as semidemocracies because the opposition cannot become the government. Indeed, those few other African states that held more than one free election after independence were also disproportionately of the dominant-party type. Among them were Somalia in the 1960s, Madagascar in the 1960s and early 1970s, and Burkina Fasso (then Upper Volta) in the early 1970s. (All of these regimes were ended by military coups.)

A South African dominant-party quasi-democracy, however, would not be likely to endure. Leave aside the possibility of military intervention. As we have seen, the reason Senegal's system endures is that opposition is compartmentalized. Confined especially to the Casamance region, it is not likely to spread. Botswana's opposition is ascriptively limited—largely to the non-Tswana Kalanga and Bayei groups and to dissident Tswana subgroups, notably the Bangwaketse. Ascriptive minorities cannot become majorities, so elections are safe. It might be thought that Inkatha support, limited to Natal Zulu, creates a similar situation, but it does not. There would be a continuing struggle for the allegiance of the Zulu. Inkatha claims, based on the desire to recapture a glorious Zulu past, would be ideologically jarring to the ANC and, if successful, potentially dangerous to the regime because of their demonstration effect for other ethnic groups. But even if all this were not so, the challenge from an unrestricted PAC and AZAPO, whose support is thus far not confined to any single ethnic group, would be potentially very dangerous. The danger is especially credible, since support for the various Black movements already has a history of fluctuation.[83]

The short of the matter is that a South African dominant-party regime would not resemble that of Senegal, Botswana, or the Gambia, and it certainly would not resemble the fluid, dominant-party systems


of India and Israel from the late 1940s to the late 1970s.[84] The South African version would be disinclined to tolerate what it saw as mortal competition from other Black parties. Such a regime, therefore, would be very likely to declare a single-party state, outlawing the opposition. As I said earlier, this is the way single-party regimes originally came into being in states such as Kenya, Guinea, Chad, Uganda, Zambia, Sierra Leone, Mali, Burkina Fasso, and Mauritania. Such regimes do not wait until the competition actually threatens to become a majority. Volatility of support, not really present in Senegal, Botswana, and the Gambia—especially volatility coupled with rising ethnic tension—is quite enough to induce the declaration of a one-party state. In South Africa, these concerns would be particularly exigent if the tension were not merely intra-African but also involved the well-armed, extreme right-wing White organizations that grew increasingly numerous, militant, and menacing after the ANC was legalized.

The Transition Process and the Changing Structure of Incentives

There is a more promising but still very uncertain vision of durable democratization. What makes it promising is that it is also solidly anchored in party interests. The central feature of such a vision is that incentives can change, and change dramatically, as the process of democratization is under way. The decision by each actor to commit or not to commit itself to the process induces reactions on the part of every other actor. Before long, a different spectrum, with different alignments, is visible, and a new structure of opportunities and constraints can present itself. The outcome is not assured by any means, but new probabilities can be assigned to some of the possibilities.

The matter is best put concretely by returning to the process in South Africa. The early phases of prenegotiations, begun in 1990, resulted in surprising collaboration between the ANC and the National Party (NP) government, which quickly discovered that they had some common interests. Both were opposed by potentially strong rivals who hoped for the failure of the negotiations. Consequently, the partners developed a common interest in making the negotiations succeed. Both also stood


to gain from order and lose from disorder, although the ANC's interest was in demonstrating its ability to create order or, if necessary, disorder, whereas the NP government's interest was in showing that its control was intact. As a result, they found themselves collaborating to some extent in the maintenance of order. So quickly were these common interests acted upon that close observers suggested, with some hyperbole, that there already was a de facto ANC-NP coalition government, that the negotiations were really about how to finalize their relationship for the future, or at least that the ANC and NP would be the main pillars in a new regime.[85]

One conception of such a relationship, well described by Hennie Kotzé,[86] involves an ANC-NP joint executive, dominating parliament and avoiding the full implications of majority rule for quite some time. This model depends on the ANC's possible desire to escape the dysfunctional demands of some of its more activist constituents for establishment of a full-blown—and unproductive—socialist state. On this view, electoral politics would bind the ANC to restructure the economy in ways that would actually prevent it from delivering on its promises for Black welfare. As a result, the two parties would agree to have multiracial government but without democracy.

Such a model depends on some very doubtful assumptions, and it is not the only model of an emerging ANC-NP coalition. It is just as easy to envision such a relationship emerging out of a more open process of constitutional consultation. This is the sort of process considered here.

Now if indeed an ANC-NP coalition were consummated in the process of moving toward electoral politics based on universal suffrage, this would produce major repercussions in the rest of the political system. On the White side, the Democratic Party would favor the coalition and find ways to join it, but parties to the right of the NP would either benefit from the electoral support of Whites unprepared for an interracial coalition or would withdraw from electoral politics into armed resistance.[87] On the Black side, determined to remain a player, Inkatha


would feel obliged to solidify its support among Natal Zulu. This is exactly what it did after the unbanning of the ANC. In a speech to Zulu chiefs, the king of the Zulu became an Inkatha partisan, accusing the ANC of being anti-Zulu: "They want no proud Zulus left. They only want subservient black Africans who say, 'yes sir, no sir, anything you like sir,' to the ANC."[88] As indicated in Chapter 2, the inception of inclusive democracy will promote the acceleration of intra-African ethnic conflict. On the other hand, the PAC and AZAPO, rejecting the ANC's interracial conciliation, would also profit.[89]

The White and Black organizations flanking the ANC and the NP would thus provide mirror-image responses. Without any doubt, they would have some ability to siphon off support, based on sellout arguments. Perhaps, in the aggregate, the ANC might suffer relatively more electoral erosion than the NP, especially if parts of the White opposition stay out of the electoral game while parts of the Black opposition join the game. It would be possible for ANC factions to take different positions on the linkage with the National Party and for the dissenters to find the Black opposition preferable to a conciliatory ANC.

The system would then consist of three differentiated sectors: a moderate, interracial middle, flanked by racially exclusive extremes. The relations of the three incipient sectors are already dynamic. As White politics polarizes, as the Conservative Party is pulled to an extreme position, and as armed White vigilante organizations threaten violence, the PAC and AZAPO are furnished further evidence for their contention that negotiations are premature and futile. With each act of moderation by the ANC and the NP, the extremes are provoked against the middle. Negotiations create support for both flanks. Each sector rebounds off the others.

If this dynamic keeps up, it has the potential to drive the middle partners closer together. By drawing the partners closer together, the extremes do the work of the moderate middle. Especially if the extreme Black flank grows in support and is ultimately willing to test its rejection of accommodation at the polls, it will then be impossible for any actor to gain electoral power alone. For this purpose, an ANC-NP coalition would be essential, as it might also be essential to the NP in its


competition with the White flank. In these competitions, each partner—perhaps joined by Coloured and Indian partners—could help the other out at the crucial margin needed for victory. There would be an inter-dependent middle.

Such help is not farfetched on either side. It is worth remembering that Black survey respondents have often given surprisingly large percentages of support to White politicians like de Klerk. (De Klerk's approval ratings among Blacks soared after the legalization of the extraparliamentary opposition in 1990.) Now such dispositions would be underpinned by genuine compromise. And it goes practically without saying that such a coalition would have a strong interest in preferring vote-pooling electoral systems, which give an advantage to parties or coalitions able to appeal across racial or ethnic lines—although, given the threefold division of the spectrum, such a coalition might prevail under a number of electoral systems. (For example, it might even win some seats with pluralities in three-way contests.) Moreover, mutual help may extend beyond elections. If, for example, the armed White organizations endanger the new democracy, suppression by a largely Black government risks a race war. Suppression by a government with a strong White component does not.

This would not be the first time a three-sector party spectrum developed in a divided society. Where incentives to compromise exist in a society with two main groups, the accommodating middle can be set off from the respective extremist flanks, which, by their continuing threat, provide the middle with its raison d'être. In fact, this is the way the Malaysian party system developed, partitioned as it was between Malay and non-Malay sectors, with a coalition of the two in the middle, fending off claims from both flanks. It was a system conducive to both democracy and interethnic conflict reduction.

The differentiation of the spectrum in Malaysia was initially the product of the exigencies of electoral politics, for the coalition came into being to contest a difficult election, but it was then solidified by the need to negotiate a constitution.[90] Electoral politics locked the coalition in place between the extremes, because the single slate the coalition put up made clear its commitment to moderation. The negotiation bonded the participants to each other. In South Africa, the negotiation might precede the election, but an electoral component, with rewards for mu-


tual cooperation, is indispensable. Even with an electoral component, a South African coalition, operating in a polarized society, might not survive.

This is a speculative, but not wholly improbable course for the parties to take in South Africa. It may not be the only course conducive to eliciting democratic commitments from aspiring hegemons. I mention it, not to demonstrate its inevitability or even its probability, for there are a great many stumbling blocks along the way, but to reveal its plausibility and also to display the mechanism that underlies it. That mechanism is perhaps best described as transformation of the structure of incentives as the result of differential commitment of various parties to the initial democratization process.

Since this is not the only course by which hegemony can be abandoned, the point deserves to be put more generally. Di Palma speaks of political actors who "trespass, intentionally or not, into new behavior (for instance, holding free or nearly free elections)," and he goes on to add that "the first and contingent choice of trespassing can induce the trespassers to yet other behavioral commitments, with further unanticipated consequences of their own. In this way, actors come to comply with the results of actions that they had taken earlier in the process with other intents and expectations."[91] This is a felicitous, if indeterminate, way of describing the larger democratic entailments that sometimes flow from ostensibly more limited commitments.

These entailments are recurrent phenomena. The Malaysian coalition itself illustrates this, for the desire to win a single set of elections preempted for many years a Malay claim to untrammeled hegemony in the polity. Parties committed to hegemony were confined to the flanks. With the passage of time—shades of habituation!—"one-race government" or an undemocratic regime came to be regarded in Malaysia as a great leap into a very dangerous unknown. The temptations, often great, were resisted.

A comparable episode blunted Hausa claims to hegemony in Nigeria in 1967. A Hausa commitment to restructure the federal system was secured, and it resulted in a new balance of power.

In the early 1960s, Nigeria, it will be recalled, was divided into three main regions. The Hausa dominated the largest of them, using it also as a springboard to pursue hegemony at the center. In 1966, however, things went very wrong in Nigeria. The so-called Ibo coup of January


was followed by terrible anti-Ibo violence in May and by a bloody northern countercoup in July. The July coup, however, involved two groups of northern officers: Hausa, mainly Muslim, and "Middle Belters," mainly Christian. When the attempted Biafra secession was imminent in 1967, the unity of the north had to be consolidated if the impending war were to be won. This required fulfillment of the aspirations of northern minorities for separate states, to be carved out of the Northern Region. Simultaneously, the Ibo secessionists would then be confronted with claims from their own dissident minorities aiming to control the new states to be created in the East. The regions were thus divided into twelve states (six of them in the north), later into nineteen and then into twenty-one.[92] Without the undivided Northern Region as an artificially powerful base, the Hausa could no longer dominate Nigeria, for reasons I have explained in Chapter 6. By these means, then, a Hausa quest for hegemony was turned into something altogether more benign and cooperative.

The new states were not reversible, and neither was the decline of Hausa hegemony. A commitment to power sharing was a by-product of a commitment to a unified Nigeria, threatened by the civil war. Equally, the three-sector spectrum could not soon be altered in Malaysia, and so commitment to accommodation there was a by-product of initial and quite idiosyncratic electoral exigencies. In each case, what could not be done directly was done, with equal effect, indirectly. In each case, an insecure future prompted risk-averse behavior that inadvertently exchanged hegemonic possibilities for immediate gains of a qualitatively different sort. In each case, potential hegemons were enticed into becoming cooperators and democrats in spite of themselves. Not deterrence, not learning, not even reciprocity—because hegemony was not overtly traded but lost as a by-product of a different transaction—explains the change. The entailments of the transaction hemmed the parties into a new structure of constraints.

That they find themselves in this position does not mean that the coalition partners like it. In Malaysia, many Malay politicians in the coalition came to regret the compromises it required, and they took every opportunity to pursue narrow ethnic interests. If the Malaysian case is any guide, the center-versus-flanks configuration makes for a most imperfectly realized democracy. Something similar could very eas-


ily appear in South Africa—which is another reason for reinforcing happenstance commitments with institutions appropriate for democracy in a divided society.

There is one implication of the Malaysian model in particular that bears on the tasks inherent in the transition to democracy. There is some uncertainty about whether everyone needs to be brought along on the same bandwagon. Di Palma suggests that there needs to be "one coalition of consent for democracy," a coalition "reaching the peripheries of the political spectrum."[93] The Nigerian experience in 1978 suggests that this is possible after a disaster, such as a civil war. But the Malaysian example and the hypothetical South African course that has been described both render this approach doubtful as a general proposition in divided societies. If democratic commitments arise out of a repartition of the political spectrum into three, then it is the center, struggling against the extremes, that carries the consent. The extremes, rejecting the compromises of the center, are essential to the power of the new incentives, but they are assuredly not part of an inclusive coalition. That renders the task of governing much more difficult than it might otherwise be, but governing a divided society is always a precarious venture. At least it is comforting to know that not everyone needs to be brought along for an effective change to occur. If everyone needed to join the bandwagon, change in a divided country like South Africa would wait forever.

No doubt, as the process moves along, others, initially reluctant, may well be coopted into participating in the democratic arrangements, lest they be left behind. One could easily imagine the Conservatives, the PAC, and AZAPO participating as electorally oriented parties on the flanks. Yet, that would still be consistent with their opposing the institutions and the compromises that underpin them. In a spectrum divided as I have hypothesized, the actors on the flanks cannot concede the legitimacy of the arrangements of the middle.[94] I do not know of any severely divided society, including those functioning along more or less democratic lines, in which the political system enjoys a large measure of broadly distributed legitimacy.

I said earlier that the belief and learning paradigm would not explain the differentiation of the middle that I have depicted. If the process


solidifies, there will be learning later, as participants protect what they have created. What they learn during the initial course of the transition process, however, are no big lessons about the dangers of conflict or the rich rewards of cooperation. Continuing to conduct themselves in accordance with self-interest, all they "learn" is that their self-interest has changed incrementally—hence the minimalist quality of what might be required for a democratic commitment.

Since the process I have postulated depends on the reactions of a variety of actors to a series of initiatives, the process can scarcely be guided step by step. Nevertheless, the situation is amenable to some deliberate action to alter the unfavorable incentives with which the transition begins. Differentiation of the players depends upon their having an invitation to negotiate, to which they can respond in one way or another. In a sense, everyone has been asked a single question: Who wants to live in the same polity, on more or less equal terms? By giving different answers, the participants have changed the structure of the situation confronting each of them. By starting the process, a South African government or comparable actor can initiate the differentiation of the spectrum at both ends. Then those in the middle can take deliberate steps to solidify their coalition by making their commitments public. Predictably, this generates the flank reactions that give the spectrum its clear, emerging configuration. Central among the coalition's commitments are the new constitutional rules, which need to be crafted carefully, with an eye on both intergroup accommodation and inclusive democratic outcomes. Thereafter, the need to justify the constitutional product can and should be a new source of mutual democratic commitment. As I shall suggest in a moment, the rejection of the extremes can be helpful here, too.

This, then, is a process that can, in part, be shaped to make it more promising, but it can still be derailed at many points. The possibilities for thwarting democratic evolution in South Africa are multiple. The conflict between emergent Xhosa and Zulu ethnic claims, the reluctance of any opposition movement to live in peaceful coexistence locally with any other, and the threat of violence from several sources comprise only some of the dangers.[95]

If, despite the dangers, an interdependent center emerges, it can still make hegemony recede as a strategy and democratic power sharing ap-


pear more promising. In particular, these developments may create interests that are congruent with the successful operation of the political institutions most conducive to democratic conflict reduction, especially electoral institutions. Democracy then has, for the first time, a reasonable chance—certainly nothing more, but perhaps nothing less either.

Improving the Odds:
Constitution Making and Democratic Commitment

Timing is a key issue in any transition to democracy. The advantages of gradualism—in permitting the growth of new elites, of interests with a stake in democracy, of forces that can check each other, of new expectations and modes of cooperation—have been stated repeatedly.[96] In general, the advantages of gradualism—many of which, restated, actually come back to deterrence, belief, reciprocity, and perhaps especially habituation—cannot be achieved in the still short time frame of any realistic delay. On the other hand, following from what I have just argued, it would be wise to move along while the incentive structures of the actors are pro-democratic. Delay may unintentionally signal retrenchment of the democratic project. An optimal strategy therefore aims to capture the benefits of speed and decisiveness rather than the longer-term benefits of gradual change.

With two important qualifications, however. Both derive from the fact that, if the transition is to entrench interests supportive of democratic competition, neither it nor the negotiation can be regarded merely as a process, a stage to be passed through. First, since the substance of the actual rules that are adopted to govern a divided society is a matter of great importance, the process must be subordinate to getting the substance right.[97] A good process and a good negotiating atmosphere are not substitutes for good institutions. Second, the process by which agreement is reached has a latent function for the institutions that emerge from it. Along the way, participants may become more or less solidly committed to arrangements they shape. Commitment is not a foregone conclusion. They may also become committed to each other, across


cleavage lines, if they are then obliged to defend their product against attack from outsiders or dissenting insiders. Both commitments are more likely if what has been produced is a synthesis, not a swap. If a party exchanges one thing for another, that party is generally committed to what was received, rather than to what was given. Commitment to the whole product is preferable, and that is, of course, a further argument for a social contract, rather than an ordinary exchange.[98]

Several implications follow. Getting the institutions right will take time, and the social contract mode is more difficult than trading horses. Trying to convince doubters is also time-consuming. All of these aims are more likely to be achieved in a multilateral constitutional convention than in a bilateral negotiation, but consummating the arrangements becomes more difficult as participants are added.[99] Yet, the credibility of the product as a constitution is greater if it cannot be made to look like a deal that benefits those few parties who happened to be in on it. The process therefore needs to be open to wider participation, but some may decline to participate, and the end product need not embrace every shade of opinion.[100]

The skepticism that would attend the efforts of a narrow band of participants to craft an agreement could be well founded. Parties to a negotiation may be able to consummate a deal, once they sit down, but there is no assurance that what suits their interests at the moment will be conducive to the democratic stability of a severely divided society over the long term. I indicated that the middle of the South African spectrum might benefit from vote-pooling electoral systems, but the partners might or might not prefer federalism and presidentialism. Whatever their preferences, it remains true that a severely divided society needs a heavy dose—on the engineering analogy, even a redundant


dose—of institutions laden with incentives to accommodation. Malaysia did not have the panoply of such institutions. Consequently, both its democracy and its interethnic accommodation have experienced slippage over time. Negotiation inevitably requires both trades and tradeoffs. (That is what the theorists of negotiation like about it: compared to the hostility and stalemate of the past, any horse-trading is an improvement.) A possible result, which needs to be countered deliberately, is a conglomerate set of institutions that can work against each other in unfortunate ways, as those of Nigeria did in the Second Republic. An institutional patchwork is a likely product of an agreement reached by a few participants.

Beyond that, the deal that is done by a narrower than necessary band of participants invites a special kind of opposition. Besides opposing the substance of the institutions, or even the need for compromise, this opposition could extend, root and branch, to the procedure that produced a private accommodation masquerading as public policy. There is, then, no escaping the possible disjunction between the institutions that are needed and the procedures that may be called upon to bring them into being, especially if the political spectrum is partitioned in the way I have described, if the participation of the extremes is grudging or absent, and if the middle parties are the moving forces of democratization. There is a need for caution at this point.

This is not to denigrate the model I have advanced of a moderate middle, drawn closer together by the opposition of the extremes. It needs to be reiterated that this model is functional in a severely divided society. But the model invites oligopolistic constitutional procedures that can delegitimize its constructive work.

That the process itself can produce some especially valuable side benefits needs to be underscored. The American Constitutional Convention, with its innumerable compromises, fostered an enduring political culture of accommodation.101 Despite the reservations urged against the document, the ratification process, carried on separately in each state, produced a degree of constitutional convergence and commitment that Americans had not previously displayed toward their institutions.102 These processes were not rapid; they were deliberate. The pro-


cess in Malaysia also drove coalition leaders of all ethnic groups closer together. They drew the lesson that many of their differences could be accommodated in a spirit of good will, and they joined to defend the arrangements they reached against forces that attacked them from both sides.103 The result was an enhanced degree of intimacy between the partners that supported the durability of the coalition. Later they were able to argue that what differentiated them from the extremes was their realism and their commitment to compromise. Their departure from hegemony became ideologized. If a speedy conclusion to constitutional negotiations risks losing such benefits, then speed is not an unmitigated virtue.

After a democratic regime begins its work, it will not be possible to sit supine. Even if there is a rush of formerly reluctant participants to take part in the new arrangements, the democratic game will not necessarily be secure. That is why the bonds that grow up among the founding generation are so important, as is the rapid use of the accommodative mechanisms of the new arrangements. With their repeated use over time, alternatives should look less and less plausible. The probability may then recede that one person, one vote, one value, and one state will degenerate into only one legal party and one last election.


previous chapter
Chapter Seven— The Transition to Democracy and the Problem of Hegemony
next section