let
600 Ethnic Groups in Confj
manage ethnic conflict, rather than to eradicate it or to aim at either a massive transfer of loyalties or the achievement of some consensus. They involve living with ethnic differences and not moving beyond them. All of these measures fall within the domain of political engineering. Not learning theory, but the theory of political incentives, inspires these more limited measures. What we shall see, above all, is that there is much more scope for constructive policy innovation in the area of ethnic conflict than policymakers in divided societies have generally acknowledged. Ethnic problems are intractable, but they are not altogether without hope.
CHAPTER FIFTEEN
Structural Techniques to Reduce Ethnic Conflict
The time is propitious for a preliminary assessment of structural techniques to reduce ethnic conflict. In recent years, a number of countries have turned to such techniques in the quest for measures to promote accommodation. Some results of innovations in federalism, regional autonomy, and electoral systems are now available. In yet other cases, territorial and electoral arrangements adopted for purposes not related to ethnicity nevertheless have had a significant impact on ethnic conflict. Consequently, there is some accumulated evidence on which to rest judgments of efficacy.
What formal institutions do is to structure incentives for political behavior of one kind or another. But they have a more powerful influence on some incentives than on others. Broad judgments are less helpful here than a searching examination of the precise effects of specific innovations. This examination consists of two parts. The first deals with territorial devolution, beginning with the far-reaching consequences of the new Nigerian federalism. The second considers the impact of an array of electoral innovations.
It has often been said that fundamental conflicts cannot be bridged by constitution-writing. No doubt there is truth in this observation, but it is a half-truth. Where there is some determination to play by the rules, the rules can restructure the system so that the game itself changes.
FEDERALISM, REGIONAL AUTONOMY, AND CONFLICT REDUCTION
In spite of the store of human experience with the relation of territory to ethnicity,1 few are the practitioners or observers who could give confi-
1. See generally lvo Duchacek, ed., "Federalism and Ethnicity," Publius 7 (Fall 1977).
602
Ethnic Groups in Conflict
dent advice on the key questions of federalism and regional autonomy:' Under what circumstances is the creation of a separate state or region likely to forestall or encourage secession? What is the optimal form and scope such arrangements should take? How many Units should there be, and should they be ethnically homogeneous or heterogeneous? What powers should be devolved upon them and what controls retained? The skillful division of authority between regions or states and a center has the potential to reduce conflict, but there is little more than dogma available about the utility of federalism; and even then the dogmas are equally divided between those who assume that territorial boundaries should follow ethnic boundaries and those who counsel that they should crosscut them.2 The occasional assumption of policymakers who contemplate federal solutions—in the 1985 Cyprus negotiations, for example—has been that federalism is a lesser version of partition and, as such, requires homogeneous states, one per group. Such assumptions should not be regarded as the last word.
FEDERALISM: THE NIGERIAN EVIDENCE
Every so often, political systems create quasi-experimental conditions in which propositions are inadvertently tested. These conditions are never as controlled as genuine experiments, but the propositions are generally more significant than those that can be tested in experiments. Nigeria
2. For the former view, see Ronald L. Watts, Multicultural Societies and Federalism (Studies of the Royal Commission on Bilingualism and Biculturalism, no. 8; Ottawa: Information Canada, 1970), 34; lvo D. Duchacek, "Antagonistic Cooperation: Territorial and Ethnic Communities," Publius 7 (Fall 1977): 3-29, at 13; Kenneth C. Wheare, "Federalism and the Making of Nations," in Arthur W. Macmahon, ed., Federalism, Mature and Emergent (Garden City, N.Y.: Doubleday, 1955), 32. See also Arend Lijphart, Democracy in Plural Societies (New Haven: Yale Univ. Press, 1977), 42-43, 163, 193; Max Beloff, "The 'Federal Solution' in Its Application to Europe, Asia, and Africa," Political Studies 1 (June 1953): 114-31. For the latter view, see Seymour Martin Lipset, Political Man: The Social Bases of Politics (Garden City, N.Y.: Doubleday, 1960), 91-92.
For purposes of this discussion, the precise differences between federalism and regional autonomy are not of great moment. In some cases, such as the Sudan, regional autonomy was accorded to particular regions by otherwise unitary governments. In others, such as Sri Lanka, the grant of autonomy to one region would not be feasible unless the same arrangements were made for all regions. The line between federal and unitary states is elusive, but, in substantial measure, the distinction turns on whether the relation between the center and the component units is regarded as having an intrinsically constitutional character, often exemplified by representation based on the units in a separate legislative chamber at the center. Whereas regional autonomy necessitates only devolution, and regional decisions are sometimes subject to a central veto, federalism generally entails some degree of independence on the part of the constituent units. Cf. Claire Palley, Constitutional Law and Minorities (London: Minority Rights Group, Report no. 36,1978), 13-14.
Structural Techniques to Reduce Conflict 603
provides a far-reaching test of the impact of federalism on ethnic conflict.3
With two quite differently designed federal systems, separated from each other by thirteen years of military rule, Nigerian federalism affords the best evidence available on the varying impact of federalism under civilian rule. The long period of military government produced greater discontinuity between the two sets of federal arrangements than is ordinarily present when federal systems are altered. For this reason, causal relationships between the new federal arrangements and the subsequent behavior of civilian politicians and electorates are fairly clear-cut. These changes in political behavior stand out in especially sharp relief, because some continuities in behavior, harking back to the period before military rule, are also evident. Despite inevitable rough edges in the quasi-exper-iment, despite the problems inherent in historical comparisons, despite the short duration of the Second Nigerian Republic (1979—83)—despite all of this, the Nigerian evidence shows that federalism can either exacerbate or mitigate ethnic conflict. Much depends on the number of component states in a federation, their boundaries, and their ethnic composition. Particularly important is the relationship of ethnic group distribution to the distribution of states.
The First Nigerian Republic (1960—66) consisted of three main regions, each of them controlled by a single ethnic majority (and a party representing it) which used its control of the region to struggle furiously for power at the center. The dominance of the three major groups in their regions weakened the representation of minorities by opposition parties. Patronage, coercion, and the apportionment of seats worked together to overrepresent the regional majorities. Most overrepresented were the Hausa-Fulani, who, with little more than half of the North's population, held nearly three-quarters of the regional assembly seats from 1961 to 1965.4 The Northern People's Congress, dominated by Hausa-Fulani and centered in the traditionally ascendant Sokoto emirate, won the vast majority of federal seats in the North and, as we saw in Chapter 9, was soon in a position to control parliament in Lagos. That control was challenged by Yoruba and Ibo parties that had also used their regions as staging areas for the battle at the center. Under this federal system, the
3. For a discussion of such "natural experiments," see Richard A. Brody and Charles N. Brownstein, "Experimentation and Simulation," in Fred I. Greenstein and Nelson W. Polsby, eds., Handbook of Political Science, vol. 7 (Reading, Mass.: Addison-Wesley, 1973), 218.
4. C. S. Whitaker, Jr., The Politics of Tradition: Continuity and Change in Northern Nigeria, 1946-1966 (Princeton: Princeton Univ. Press, 1970), 324.
604 Ethnic Groups in Conflict
power of the three main groups, who together comprised less than two; thirds of the total population, was not merely reflected at the center; it was magnified by their control of regions whose boundaries extended somewhat beyond ethnic boundaries.5
The advent of military rule and the Biafra war provided the occasion for restructuring the federal system. To counter the Biafran charge that Nigeria was dominated by the large Northern Region, to secure the loyalty of Northern minorities strongly represented in the army, and to wean the Eastern minorities away from Biafra, the regions were carved into twelve states in 1967.6 The three main groups were relegated to seven states. The dissolution of the old political parties by the military and the creation of new states together liberated the minorities from control of regionally dominant groups and paved the way for new alignments.
The new fluidity was enhanced by the creation of seven more states in 1976, before the return to civilian rule.7 Now the Hausa were spread among half a dozen states, the Yoruba among five, and the Ibo between two. The proliferation of states produced a lively state politics and a more complex—and therefore less tense—politics at the center. Both of these effects of the new states were visible when civilian politics re-emerged in 1979,8 and both effects would increase if more new states were created.
The new arrangements transferred a good deal of conflict from the all-Nigeria level to the state level. State-level conflicts have been intraeth-nic and interethnic. More than half the nineteen states have a substantial measure of ethnic heterogeneity, especially in the former regional minority areas. In such states, there has been tension over the allocation of civil service positions and other benefits to various groups and areas of the state. Disappointed groups have in several cases demanded creation of their own separate states. The Ibo and Yoruba states have much less
5. This was no longer true of the Yoruba after a fourth, largely non-Yoruba region was carved out of the Yoruba Western Region in 1963. Thereafter, the West, save for cosmopolitan Lagos, was nearly all Yoruba.
6. For a succinct analysis, see Crawford Young, The Politics of Cultural Pluralism (Madison: Univ. of Wisconsin Press, 1976), 301-08. See also Brian Smith, "Federal-State Relations in Nigeria," African Affairs 80 (July 1981): 355-78.
7. The nineteen states were intended, as were the twelve, to respond to the demands of minorities and to subdivide the large ethnic groups further. See Ali D. Yahaya, "The Creation of States," in Keith Panter-Brick, ed., Soldiers and Oil: The Political Transformation of Nigeria (London: Frank Cass, 1978), 201-23.
8. The observations that follow are based on interviews 1 conducted in Nigeria in 1978, supplemented by several additional interviews I conducted in 1980.
Structural Techniques to Reduce Conflict
605
ethnic heterogeneity, but they have experienced considerable intraethnic tension. Disputes have arisen in Imo, an Ibo state, between Bende and Owerri administrative divisions over jobs and development projects; the same sort of quarrel has occurred in Yoruba states, such as Oyo, among subgroups clustered in Ife, Ilesha, and Oshogbo. The boundaries of the Ibo and Yoruba states tend to follow—and therefore reinforce—-existing lines of subethnic cleavage. The Owerri Ibo in Imo have long been at odds with the Onitsha Ibo of Anambra, while the Oyo and Ogun Yoruba, consigned to states with those names, have disputed the apportionment of resources between their two states. The nineteen states thus created a new, lower layer of conflict-laden issues around which already existing differences crystallized, greatly reducing the previously unchallenged importance of contention at the all-Nigeria level.
This dispersal of conflict was reinforced by a distributive side effect of creating nineteen state bureaucracies. The federal and state bureaucracies are differently composed. Inevitably, the ethnic composition of state civil services tended to reflect the composition of the respective states, albeit not at all proportionately within heterogeneous states. These new bureaucratic opportunities reduced—though they by no means eliminated—the potentially explosive significance of a major ethnic issue at the center: the disproportionate representation of Yoruba, Edo, and a number of smaller groups in the central bureaucracy following the Ibo departure for Biafra.
None of this state-level conflict was sufficient in the South to prevent the recrudescence of essentially Ibo and Yoruba parties at the all-Nigeria level. In the 1979 senatorial, gubernatorial, and presidential elections, these two parties won between 74 and 95 percent of the vote in the solidly Ibo and Yoruba states. With the exception of a minority segment of Ibo who supported one of the parties with a substantial Northern base, at the all-Nigeria level of politics the subgroup loyalties of the two main Southern groups did not effectively compete with overarching group loyalties. Indeed, as the 1979 elections approached, both the Yoruba and Ibo parties grew more cohesive. Opposition to the Yoruba United Party of Nigeria (UPN) in the Yoruba Oyo and Ondo states was overcome, and Ibo flocked from other parties to the banner of the venerable Ibo nationalist Nnamdi Azikiwe when he took command of the Nigerian People's Party (NPP). Despite some changes in group cohesive-ness in the 1983 elections, both Ibo and Yoruba parties persisted.
Major changes were wrought in the alignments of minorities in the
606
Ethnic Groups in Conflict
Structural Techniques to Reduce Conflict
607
East and North. The Biafran hostilities had widened the gap between Ibo and the minorities in the former Eastern Region. Now concentrated in Rivers and Cross River states, the Eastern minorities tended, in the Second Republic, to align with Northern-based parties. Non-Muslim minorities in the North were also free to choose new alignments. The period of military rule had sharpened tensions between groups strongly represented in the armed forces—notably between Hausa and certain Middle Belt groups—particularly after the execution of Middle Belt officers implicated in the 1976 assassination of Murtala Mohammed. Now Plateau state, the heart of the Middle Belt, eschewing Northern alignments, supported the predominantly-Ibo Nigerian People's Party.
Most striking of all the changes was the new political significance accorded by the proliferation of states to cleavages within the Muslim North. These cleavages, always observable, were formerly muted by the undivided regional structure and the overriding struggle between the Muslim-majority North and the South.
The new states breathed new life into former Northern opposition parties based on ethnic and subethnic differences. Previously, it made little sense to support a party that would be consigned to futile and often punishing opposition in the Northern Region as a whole. In the federal election of 1959, the Northern People's Congress (NPC) won all but 6 of the nearly 100 Northern constituencies outside the Middle Belt.9
Now, however, electoral incentives were transformed. The pains of opposition in the undivided region became the rewards of power in the states. The creation of ten states in the former Northern Region simultaneously brought into being ten state legislatures, ten governors, and, since each state sent five senators to Lagos, fifty federal senators. A party with only minority support in the old North could now control one or more states, could gain one or more federal senate seats—for these are apportioned by territorial constituencies within states—and could have a respectable voice at the center. There had always been leaders willing to take up the cudgels of opposition in the North, but other leaders and
9. Richard L. Sklar, Nigerian Political Parties (Princeton: Princeton Univ. Press, 1963), 338. Properly modified for ethnic differences, the situation of Northern Nigeria under the NPC was not far from that described by V. O. Key, Jr., for one-party Virginia under the Byrd machine. For opposing parties to emerge, Key notes, "each party must, almost of necessity, have a territorial stronghold in which it can win legislative election and control local governments. . . . The punitive powers of the organization, through its control of the perquisites of local officials and its ability to obstruct local bills in the legislature, can discourage competing factions territorially segregated." Southern Politics in State and Nation (New York: Vintage Books, 1949), 33.
most voters had responded more sensibly to the logic of the situation by supporting the NPC. The new state boundaries turned that logic around, making ethnic and subethnic loyalties at the state, rather than regional, level more important in determining party support. The new federal structure thus facilitiated the expression of Northern heterogeneity more accurately than the earlier regional structure had permitted.
Two parties in the Muslim North benefited from the new structure. One, centered in Borno, was dominated by the Kanuri, a large majority in that state but only about 10 percent of the North as a whole. The other, centered in Kano state, embodied elements of Kano resistance to Sokoto,10 Hausa resentment of Fulani overlordship, and social-class differences. Both parties had direct antecedents in the First Republic.
The Kanuri had been tied closely to the ruling Northern People's Congress in the First Republic. Some Kanuri, to be sure, had organized the Bornu Youth Movement (BYM), an explicitly Kanuri party. But the NPC carefully nurtured Kanuri support by a combination of rewards and pressure. The Bornu Youth Movement fought back, and interparty conflict, sometimes overlapping Hausa-Kanuri conflict, was often bitter. Nonetheless, the BYM won few elections and was confined largely to urban areas.11
In 1976, however, the Kanuri were awarded their own Borno state. They responded by supporting their own Great Nigeria People's Party (GNPP) to govern it.12 In 1979, the state assembly, the governorship, and four of Borno's five senate seats were won by this party; the fifth was narrowly lost in a triangular race. The GNPP's strength also spilled into adjacent areas in the north of neighboring Gongola state. With some historical ties to the old Bornu empire, Gongola gave the GNPP a plurality of its state assembly seats (enabling it to form a fragile minority government in that state), as well as the governorship and two senate seats.
The Kanuri had fought and repelled Fulani armies in the nineteenth century. It is not surprising that, once it became possible to control their
10. Cf. Whitaker, The Politics of Tradition, 279-82; John N. Paden, "Islam, Constitutional Change, and Politics in Nigeria" (unpublished paper, Northwestern Univ., Oct. 1979), 4.
11. Sklar, Nigerian Political Parties, 339-44; Whitaker, The Politics of Tradition, 386.
12. The GNPP was not intended to be solely a Kanuri party, for it was earlier linked to a party with Ibo support, as explained below. But that does not detract from the fact that it was assuredly not linked to the successor to the NPC.
It should be noted that the spelling of Bornu has now been altered to Borno.
608
Ethnic Groups in Confl
own state government, most Kanuri would leave their ties to a Hausa-Fulani party behind them. The GNPP, which was led, significantly, by a former NPC cabinet minister, Waziri Ibrahim, could easily be viewed as the Bornu Youth Movement writ large.13 Ibrahim, however, proved maladroit at managing intraparty affairs. By the time of the 1983 elections, the GNPP was badly factionalized. Many of its leaders defected to other parties. In the presidential election, Ibrahim received fewer than half the Borno votes he had received in 1979, and the state fell to the National Party of Nigeria (NPN) in the later gubernatorial and federal legislative rounds as well.
Despite Ibrahim's quixotic quest for national power and his eventual rejection by the Kanuri, the general point remains: carving the old North into ten states created ethno-political opportunities foreclosed by the earlier system. The proof lies in another Northern state, Kano, whose political independence in 1983, as in 1979, survived the death of the founder of its ruling party.
Aminu Kano's People's Redemption Party (PRP) had considerable success in 1979 in the Hausa-Fulani state of Kano and in neighboring Kaduna state. The PRP was the reincarnation of the old, unsuccessful Northern Elements Progressive Union (NEPU), also led by Aminu Kano. In the First Republic, NEPU could be suppressed by the NPC's judicious use of regional patronage, by accusations that NEPU was dividing the North in the great struggle to control Nigeria, and by the taint of its affiliation with the leading Ibo party.14 In the Second Republic, however, patronage was controlled at the state level, it was more difficult to see the North-South struggle as all-encompassing, and the other Northern parties were the strangers in Kano compared to the PRP. In 1979, Aminu Kano's party captured the Kano state assembly, the governorship, and all five senate seats by overwhelming majorities. In neighboring Kaduna, the PRP won two senate seats, the governorship, and a strong minority position in the state assembly. It was a far cry from the lean years of NEPU's fruitless opposition, during which, in six regional and federal elections between 1956 and 1964, the party had won a grand total of eighteen seats. In the last Northern regional elections, in 1961, NEPU
13. So uncanny, in fact, was the resemblance that the former inferparty, interethnic violence has reappeared in Borno; and, in the 1979 election, the traditional ruler of the Bornu emirate, the Shehu, threw his support to the NPN just as he had earlier supported the NPC over the Bornu Youth Movement.
14. Whitaker, The Politics of Tradition, 385, 410.
Structural Techniques to Reduce Conflict
609
had won one seat to the NPC's 156.15 While factionalism hurt the PRP in Kaduna in 1983, the party did very well again in Kano, winning the governorship, nearly every federal and state seat, and the presidential vote in the state.
Many of the forces that had supported the NPC in the First Republic were drawn to the NPN in the Second. In 1979, the NPN did well in many traditional areas of NPC strength—indeed, in practically all except those areas where the GNPP and PRP showed the strength denied to their antecedent parties in the First Republic. This limitation, however, was enough to crimp the Northern support of the NPN, so that it won, for example, only half the governorships in states of the former Northern Region. In 1983, however, in an election marred by fraud, the NPN won all the Northern governorships except Kano.
Each of the five major parties of the Second Republic was thus recognizable as a somewhat altered version of one or more of the parties of the First Republic. This was most abundantly clear for the GNPP and the PRP. The Yoruba UPN was a recrudescence of the Action Group but without the Action Group's allies outside Yorubaland. The mainly-Ibo NPP inherited most of the support of the Ibo NCNC, but with the addition of support in Plateau state that would earlier have gone to the old United Middle Belt Congress. In the First Republic, the UMBC had been aligned with the Action Group. The NPN, with the same following the NPC had in many Hausa-Fulani areas of the North, also developed a strong following in the Ibo state of Anambra and in the Eastern minority areas. Much of the latter support had formerly been linked to the Action Group. These elements of party continuity attest to the persistence of ethnic forces in Nigeria, even in the face of certain changed ethnic alignments and of NPN strength outside the Muslim North.
The main impact of the new federal structure on party politics was, then, not to abolish ethnically based parties. To the contrary, as parties formed and re-formed in 1978-79, it was easy to discern the sorts of mergers and splits that are so characteristic of the evolution of ethnic parties in divided societies. The GNPP, for example, was originally part of the now mainly-Ibo NPP. It split off when Azikiwe took the NPP leadership, depriving Waziri Ibrahim of a chance to be the NPP presidential candidate. Originally, Ibrahim took some Ibo leaders with him to the GNPP; but these returned to the NPP camp after the 1979 elections,
15. Ibid., 374.
-a o
O
2 an w X H PS
o
2
2 oo -s
H M c
J Q "~
On c
M O
2 O
H
U
w
-> ~ -c
ID <
08 C <u
m oo
ON
C
ON
3 O
I
ON
ON
	
to	
	
Hoi	O IN
	II
OO	K
ON	
ON	
r"H	
I)	
	
3 o	O
	
On	II
	s
ON	—
r-H	
O   t   CO rs|
OO   7}-   OO   OO (N
Z ft.
z
ft.
ft, Z ft. £~ £. & Z ft. D Z U
U ft. ft. w
Z Z
u
ea.
s
o <
60 2
O
2:
X: OO
s >2 5 ■ »
o ^ . a. 4- m
<i> K .h:
ro 73
I—  - c
4£ I
_   V =3
id 3
y
o c* e
••   C It
3 s° g c s
Structural Techniques to Reduce Conflict
611
leaving the GNPP with no significant Southern support and the NPP with no support north of the Middle Belt. Not an end to ethnic parties but a rearrangement of the building blocks of such parties, and—in 1979—a less distorted reflection of their underlying strength, particularly in the North: these were the main consequences of the new federal structure.
The results of this in federal legislative elections for Northern constituencies are visible in Table 11, which lists parties competing in the North in the First Republic opposite their nearest successors in the Second. The table shows plainly that the difference in 1979 was the way seats Were spread among the Muslim parties of the North: the decreased strength of the NPN vis-a-vis its predecessor, the NPC, and the concomitantly increased strength of both the Kano-based PRP vis-a-vis its predecessor, NEPU, and the Kanuri-based GNPP vis-a-vis the Bornu Youth Movement.
Looking carefully across the first line of the table, depicting NPC and NPN percentages, it appears as if the 1983 results entailed a reversion to the pattern prevailing in the First Republic, whereby the party representing Hausa-Fulani was able to capture support disproportionate to its underlying population in the North. There are, however, two considerations that militate against such a clear-cut interpretation. First, two-thirds of the NPN increase from 1979 to 1983 was due to the collapse of the GNPP. Indeed, despite the GNPP's factionalization and defections, the NPN presidential candidate won both Borno and Gongola states with less than a majority—which is quite different from the overwhelming support the NPC enjoyed there in the First Republic. This rather clearly implies that a revival of Kanuri cohesion, under the aegis of the GNPP or some other party, would restore the 1979 NPN results. Second, there is something more important about the 1979 results. The first election under the new constitution was a watershed event. Since, in that election, the NPN could not count on the undivided support of the North, it proceeded to cement relationships with groups outside the North, as I shall point out very shortly. These relationships survived the change in the Northern balance of power in 1983, so that, while the NPN was a Hausa-Fulani party in the North, it was not just that in the country as a whole. The fortuitous collapse of Kanuri cohesion in 1983, rather than in 1979, is testimony to the capricious but decisive significance of sequences of behavior in determining the impact of institutional innovations.
612
Ethnic Groups in Confl
let
It may seem odd to lay so much emphasis on the electoral impact of' the new states, but it is not wide of the mark. The Nigerian experience shows that federalism can act as a kind of electoral reform, setting off one arena from another, making and unmakinglegislative majorities and minorities by adjusting the territory in which their votes are to be counted. As the new states changed electoral incentives, their effect in the North was rather like a change in electoral formula. The new states substituted something akin to proportional representation for the win-ner-take-the-whole-region formula that had exaggerated the strength of the largest regional party and of the largest ethnic group that comprised the party's support. If additional states are created, and if elections return to Nigeria, this proportionality effect should increase.
The new states reduced the overall power of the Hausa-Fulani, so it was no longer plausible for them to think of dominating the whole country. The reduced strength of the NPN in the North in 1979 gave the party a powerful incentive to do what new electoral requirements also encouraged it to do—appeal to ethnic groups outside the core area of its support and, indeed, outside the North altogether. The NPN had more success in reaching out to groups across the country than any other party did: it won the presidency and the largest number of seats in both federal houses in 1979 and in 1983. That the new states had the secondary effect of reinforcing incentives for interethnic cooperation across North-South lines is evidence of the efficacy of using territorial boundaries as an instrument of conflict reduction. Federalism, in short, can create a new framework for electoral reasoning on the part of voters and party leaders. By heightening the importance of cleavages within the North, the new framework ultimately reduced more dangerous cleavages at the all-Nigeria level.
Finally, the creation of nineteen states ranging in population from less than two million to more than eight million also created incentives for political actors to see at least a few all-Nigeria level issues in terms of competition among states, rather than among ethnic groups. Electoral issues in the Constituent Assembly that framed the new constitution occasionally arrayed large states against small states. On revenue issues, rich states have argued that revenue should be spent where it is "derived"; resource-poor states, that revenue should be apportioned by population. Interethnic alliances are notable on both sides.
The new Nigerian federal framework thus utilized all five mechanisms of conflict reduction enumerated earlier. First, the proliferation of states
Structural Techniques to Reduce Conflict
dispersed some of the conflict into more parochial forums. Second the new states provided arenas in which intraethnic conflict might also occur. Third, a result of this was to enhance the position of some political parties at the expense of others, especially in the North, paving the way for greater interethnic cooperation in the all-Nigerian arena. Fourth, as the new states fought to advance their interests, a few nonethnic issues and actors were also introduced. And, fifth, the separate state bureaucracies provided career opportunities for groups not well represented in the federal civil service. Using all these tools, the nineteen states readjusted, realigned, and complexified the Nigerian political system.
In many ways, of course, there was substantial continuity between the First and Second Republics. Yet, even in party politics, where that continuity was manifest, what is striking is the considerable importance of rather small changes in party support. To be sure, the electoral impact of the new states was abruptly terminated by the (nonethnic) military coup that occurred on the last day of 1983. Nevertheless, in the wake of all the failures of planned change in the developing world, the Nigerian arrangements had by then already demonstrated that it is possible to take deliberate action to restructure institutions so as to alter ethnic balances and alignments.
THE DESIGN OF TERRITORIAL STRUCTURES
The Nigerian experiment provides a demonstration of the importance of political context in shaping the manifestations of ethnic conflict. If the results are replicable, territorial design may prove a useful instrument of conflict reduction.
In spite of the small number of federal states in Asia and Africa, several conclusions from the Nigerian experience can be corroborated. There is already good, if limited, evidence on the ethnic impact of homogeneous and heterogeneous states, the consequences of alternative ways of drawing boundaries, and the costs of devolution in terms of ethnic conflict.
Homogeneous States
If groups are territorially separate and subethnic divisions are prominent, the case for ethnically homogeneous states is strong. (The term homogeneous, needless to say, does not preclude subethnic cleavages.) India, which moved toward "linguistic" states in the 1950s, provides abundant testimony. The classic case is Andhra Pradesh, a state of Tel-
614
Ethnic Groups in Conflict
Structural Techniques to Reduce Conflict
615
ugu-speakers previously merged with Tamil-majority Madras. The Tel-ugu movement for a separate state "assumed such an intensity that it was unimaginable that within a decade language .'. . would have to contend with other claims."16 Yet, after a separate Andhra was conceded in 1953, language conflict was superseded by-the unremitting struggle between the Kamma and Reddi castes, both Telugu, to control the state, not to mention the equally fervid struggle between the Telangana region (which joined the state in 1956) and the Coastal region of Andhra.17 Subethnic differences have also preempted politics in other monolingual Indian states. Kerala politics has revolved around the rivalry of Ezhavas, Christians, Nairs, and Muslims, a quadrilateral configuration of caste and religious conflict replicated in no other state.18 Neighboring Karna-taka, designed as a homogeneous Kannada-speaking state, "since its inception has been a silent spectator of the uninterrupted virulent race for power between Lingayats and Vokkaligars."19 Bihar has its quarrels between tribals and caste Hindus. In each case, broader territorial boundaries would probably produce broader ethnic conflict, along the lines of Tamil-Telugu conflict in pre-1953 Madras.
The complexity of Indian society has facilitated the flow of conflict in linguistically homogeneous states into subethnic channels, just as it has in Nigeria's homogeneous states. Under such circumstances, devolution of a generous share of power upon largely homogeneous federal units promises a dramatic reduction in conflict at the center. Many issues will be contested within ethnic groups, rather than between them, simply because many contested issues become state-level issues. It is difficult to infer causality from Switzerland, because it has not had intense conflict, but it has been argued that Swiss federalism, with its powerful and mainly homogeneous cantons, is effective in dampening ethnic conflict because of the sparseness of contentious issues at the confederal level of
16. Jyotirindra Das Gupta, "Ethnicity, Language Demands, and National Development in India," in Nathan Glazer and Daniel P. Moynihan, eds., Ethnicity: Theory and Experience (Cambridge: Harvard Univ. Press, 1975), 485.
17. G. N. Sharma, "Aspects of Andhra Politics," in Iqbal Narain, ed., State Politics in India (Meerut, India: Meenakshi Prakashan, 1968), 96-104.
18. Lloyd 1. Rudolph and Susanne Hoeber Rudolph, The Modernity of Tradition: Political Development in India (Chicago: Univ. of Chicago Press, 1967), 71-76; Selig Harrison, India: The Dangerous Decades (Princeton: Princeton Univ. Press, 1960), 196-99; V. K. S. Nayar, "Communal Interest Groups in Kerala," in Donald Eugene Smith, ed., South Asian Politics and Religion (Princeton: Princeton Univ. Press, 1966), chap. 8.
19. Sushi] Kumar, "Panorama of State Politics," in Narain, ed., State Politics in India,
68.
politics and the "tranquilizing effect" of compartmentalizing them.20 Whether or not this is what works in Switzerland, it certainly seems promising for countries with these attributes: serious conflict at the center, territorially separate groups, and significant subethnic divisions. It is essentially what was proposed for Nigeria by Ibo leaders on the eve of the Biafra secession: "Since the control of the Centre has been the main cause of friction and tensions between the different Regions, thereby threatening national solidarity and integrity, the distribution of functions between the Regions and the Centre should be reviewed and so arranged that only such subjects and functions as will engender the minimum of suspicion and friction among different groups are allowed in the hands of the Federal Government."21 Few states in Asia and Africa, however, would be willing to accord as much power to constituent units as Switzerland does to its cantons.
Even less generous devolution, however, promises some results. The regional autonomy agreement put into effect for the Southern Sudan in 1972 carefully limited the powers of the People's Regional Assembly it created for the South, and the agreement specified that the regional Executive Council was to be responsible to Khartoum.22 Limited though Southern authority was, it was sufficient to set in motion a contest for control of the region that brought to the fore all the intra-Southern differences that had been manifested during the civil war: between those who spent the war years in Khartoum and those who spent them in exile or in the bush, among the various Southern parties, and among the several main ethnic groups in the South. As the regional government began its work, new intra-Southern issues arose. Now it became imperative that bureaucratic appointments and expenditures not favor one group or area over another.23 The autonomy agreement seems to have
20. Hans Daalder, "On Building Consociational Nations: The Cases of the Netherlands and Switzerland," in Kenneth D. McRae, ed., Consociational Democracy: Political Accommodation in Segmented Societies (Toronto: McClelland & Stewart, 1974), 110; André Siegfried, Switzerland: A Democratic Way of Life, trans. Edward Fitzgerald (New York: Duell, Sloan & Pearce, 1950), 161.
21. Statement of Lt. Col. Ojukwu to Diplomatic Representatives of the U.K. and U.S.A., Sept. 1966, reprinted in Nigerian Crisis, 1966: Eastern Nigerian Viewpoint (En-ugu: Eastern Nigeria Ministry of Information, 1966), 42.
22. "The Addis Ababa Agreement on the Problem of South Sudan," Chaps. IV-V1, in The Grass Curtain (London) 2 (May 1972): 18-20.
23. Nelson Kasfir, "Southern Sudanese Politics Since the Addis Ababa Agreement," African Affairs 76 (Apr. 1977): 143-66; James E. Sulton, Jr., "Regional Autonomy in the Southern Sudan: A Study in Conflict Regulation" (Ph.D. diss., Johns Hopkins Univ., 1980), 200-03, 255-56, 410; Sudanow (Khartoum) 5 (June 1980): 12.
616 Ethnic Groups in Conflict
achieved some balance between North-South issues and intra-Southern issues, but the balance was undone by the controversial division of the South into three regions in 1983 and the resurgence of warfare thereafter.
The development councils adopted in Sri.Lanka in 1980 are also centrally controlled, but they, too, augur some restructuring of conflict. Elected councils were set up in every administrative district. However, members of parliament comprise a majority of members of each council, and a centrally appointed district minister sits as a member of the council's executive committee. The councils operate as local authorities in rural areas and have limited powers of taxation, but the district minister and the central government he represents retain ultimate authority.24 In order to make them more palatable to Sinhalese opinion, which fears Tamil separatism, the development councils were portrayed as mere adjuncts of the district ministers, useful for decentralizing development functions.2S In fact, they were designed to be district legislatures. At the time of adoption, some policy planners spoke of the councils as a "settlement" with the Tamils, an arrangement for "quasi-federalism," or "24 states."26 Plainly, the councils were intended to devolve power upon local authorities which, in the North and East, would be Tamil-dominated.
That, indeed, is how they were perceived by Muslim members of the commission appointed to make recommendations regarding the structure of the councils. In a concurring report, two Muslims pointedly opposed inclusion of the Muslim minority, heavily concentrated in the Eastern Province, within the jurisdiction of Tamil-dominated units; they also opposed elections to the councils.27 Their apprehension reflects the
24. By the terms of the Development Councils Act, No. 35 of 1980, sections 61-63, the district minister is empowered to remove any council member or the chairman of the council for incompetence, mismanagement, abuse of power, unlawful behavior, or default in the performance of duty, while the president of Sri Lanka is authorized to dissolve a council's executive committee if there are differences between it and the district minister or to remove executive committee members for incompetence or mismanagement. For a survey of the councils, see Bruce Matthews, "District Development Councils in Sri Lanka," Asian Survey 22 (Nov. 1982): 1117-34.
25. See, e.g., Far Eastern Economic Revieiv (Hong Kong), Aug. 17, 1979,p. 15.In an interview, however, President Jayewardene referred revealingly to district ministers as becoming "rather like the Chief Ministers who run Indian states." Ceylon Observer (Colombo), Apr. 6, 1980.
26. 1 am drawing here and elsewhere in this discussion on interviews 1 conducted with several members of the Presidential Commission on Development Councils and some leading politicians and policymakers, Sinhalese, Tamil, and Muslim, in Colombo in April 1980.
27. Report of the Presidential Commission on Development Councils, Sessional Paper No. V—1980 (Colombo: Government Publications Bureau, 1980), 83-95 (Note of Reservation by Mr. A. C. M. Ameer, Q.C., and Mr. M. A. Azeez).
Structural Techniques to Reduce Conflict
617
fact that, as Tamils are outnumbered by Sinhalese in Sri Lanka, Muslims are outnumbered by Tamils in the Eastern Province. In the first elections to the councils, held in 1981, the Tamil United Liberation Front did indeed capture control over the six councils in the North and East.
Were it not for the severe anti-Tamil riots of 1983, as well as the continuing Tamil terrorism and army reprisals, it would seem safe to conclude that the Sri Lankan devolution might portend a good deal less Sinhalese-Tamil conflict at the center, in exchange for somewhat more intra-Tamil and Tamil-Muslim conflict in the districts of the East. This last, of course, is not subethnic, for the Muslims are not Tamils (though they are Tamil-speaking), but it is certainly conflict at a lower level and with different alignments. Where groups are territorially concentrated, devolution may have utility, not because it provides "self-determination," but because, once power is devolved, it becomes somewhat more difficult to determine who the self is.
That the riots and the terrorism occurred despite the devolution is, of course, no evidence against the efficacy of devolving power. The terrorism that precipitated the riots was, as 1 shall suggest in Chapter 16, the product of long-standing Tamil frustration. What the violence shows, once again, is the importance of timing: accommodation long delayed may be accommodation ultimately denied.
Heterogeneous States
Where groups are territorially intermixed, some reduction in conflict at the center may be achieved by the creation of heterogeneous states. The Malaysian federal system illustrates several purposes that may be fulfilled by carving out states with varying degrees of heterogeneity.
First, as the new states in Nigeria showed, federalism can create political compartments in which ethnically and subethnically differentiated parties can flourish. If ethnic ratios vary from state to state, a group that is a minority at the center may be a majority in one or more states and may be in a position to rule these states, thereby mitigating its reduced influence or even exclusion at the center. This was the theory behind the unwritten arrangement that gave the Chinese a preeminent voice in the Chinese-majority state of Penang. Like reserved offices—which it practically is—this arrangement tends to exacerbate subethnic divisions and promote intraethnic party competition.
Second, state governments provide opportunities for the development of interethnic elite relations that sometimes soften ethnic hostility among politicians. In Malaysia, land is a subject reserved to the states. Chinese
618
Ethnic Groups in Conflict
Structural Techniques to Reduce Conflict
619
businessmen who need land for development or who need state approval for other projects find it advantageous to develop close relations with the Malay chief minister or with members of the state executive council. A frequent result of these close business relations is.that the Malay politicians involved become "much quieter" on ethnic issues.28
Third, though it may seem unlikely, even ethnically heterogeneous states may produce a quarantine effect for certain issues of ethnic conflict. Two of the worst outbreaks of ethnic violence in Malaysia—in Penang in 1967 and in Kuala Lumpur in 1969—were largely, though not entirely, confined to the states in which they began. The issues which precipitated the violence may have been seen as local or state issues; in the latter case, the precipitant was related to the results of the state elections. Quarantine effects seem more likely where state boundaries have some historical foundation, as they do in Malaysia, and are not merely convenient constructs. The creation of ethnically heterogeneous states may help scale down to the state level some divisive issues that might otherwise engulf the entire country.
If this is so, the benefits of heterogeneous states challenge the conventional wisdom that federalism is an apt prescription for ethnic conflict only when groups are territorially compact and therefore amenable to encapsulation in homogeneous units. In fact, the prospects for federalism are more complex than the conventional formula would indicate, for neither homogeneous nor heterogeneous units are useful under all conditions.
As 1 have noted, what makes homogeneous states useful in conflict control is the existence of lower-level cleavages that are activated in state politics. Without subethnic divisions or lower-level ethnic divisions than those that prevail in politics at the center—or with a structure that suppresses such lower-level divisions, as the old regional system did in Nigeria—homogeneous states are unlikely to reduce conflict at the center. Rather, they are likely to be, as the Nigerian regions were, springboards
28. The phrase is drawn from an interview 1 conducted with a Malay politician in Kuala Lumpur, Jan. 1968. Joseph S. Nye has pointed out that one function of corruption is sometimes to "overcome divisions in a ruling elite that might otherwise result in destructive conflict," and he specifically notes that elites based on power and on wealth sometimes have "assimilated each other" through cotrupt payments. "Corruption and Political Development: A Cost-Benefit Analysis," American Political Science Review 61 (June 1967): 417-27, at 420. In Malaysia, interethnic elite relations at the state level are greatly facilitated by the multiethnic alliance that has long ruled the country and nearly all the states. Because of this party arrangement, state-level Malay and Chinese politicians simply'must deal with each other.
to group power at the center. Especially if there are only a few units, homogeneous states on the model of the First Nigerian Republic will exacerbate rather than alleviate ethnic conflict. Consequently, not all homogeneous states are an improvement on no states at all.
On the other hand, heterogeneous states sometimes are a decided improvement on unitary government. However, as we shall see in a moment, heterogeneous states embracing territorially separate groups, each occupying a portion of the state, have some tendency to fission. In practical terms, therefore, heterogeneous states with the greatest potential for reducing ethnic conflict are those whose groups are intermixed or whose territorial boundaries have some long-standing binding force, as the Malaysian states (formerly sultanates) have.
In short, federalism is not for everybody. The federal judgment must be a differentiated and prudential one. Even so, it is safe to say that federalism or at least some devolution has conflict-reducing possibilities for many more countries than have so far contemplated it.
The Vine Art of Devolution
Whether to attempt to use homogeneous or heterogeneous states in conflict reduction, and how much power to devolve, depends on which of the underlying mechanisms of conflict reduction stands the best chance of functioning in a given environment. If intraethnic cleavages can be utilized to reduce the energy expended at the center in interethnic conflict, their availability points to homogeneous states and, as indicated earlier, a generous grant of power. Similarly, if group disparities coincident with region are so pronounced that the aim is to use federalism for distributive purposes, either within states or at the center by allocating opportunities by states, then, too, homogeneous states are called for; and if the aim is specifically to create state bureaucracies composed differently from the federal one, this argues for more state-level powers and functions. If the objective is simply to take the heat off an overheated center, then heterogeneous states may accomplish this and in the process help quarantine some conflict-producing issues that might otherwise be carried to the center. If, however, federalism is to foster interethnic cooperation, there are two routes to this. One is the Nigerian, which, using homogeneous and heterogeneous states, at first whittled down the power of the largest Northern group, thereby heightening incentives for interethnic cooperation at the federal level. The other route is the Malaysian, which proliferates the occasions for interethnic cooperation at the state
620
Ethnic Groups in Conflict
Structural Techniques to Reduce Conflict
621
level in mainly heterogeneous states, so that state politicians who find their way to the center have already had experience in dealing with leaders of other groups.
Generally, if states are heterogeneous, this points toward more cautious devolution, so as not to jeopardize interethnic cooperation that can be built up at the state level. Moreover, if significant goods can be obtained at the center, the possibility increases that some issues at the center will be defined in terms of state rather than ethnic interest, and competition for those goods will involve one ethnically heterogeneous unit against another. So again, if states are heterogeneous, a more powerful center is advisable.
More often than not, perhaps, homogeneous states will be indicated, but the availability of various mechanisms of conflict reduction—hence the attractiveness of one kind of unit or another—depends on the configuration of divisive issues, as well as the territorial distribution and internal structure of the groups. It needs to be underscored, however, that many regimes will be reluctant to devolve power on homogeneous regions that have exhibited separatist inclinations, and many more will be reluctant to weaken central authority by significant grants of power. More about these qualms shortly.
How many units are optimal for conflict reduction in a federal system is unanswerable for more than one reason. First of all, it depends on the size of groups relative to the state and relative to each other. The Hausa imbalance in Nigeria clearly called for a significant number of states if ethnic and subethnic divisions in the North were to be brought to bear in the creation of incentives to interethnic cooperation. The Southern Sudan did not present the same problem, and a single Southern region was sufficient to bring subethnicity into play. Equally important, it is difficult to forecast when an exercise in state creation will prove enduring. There is usually room for subsequent pressure to alter the number and boundaries of states. India and Nigeria have been through several waves of proliferating states, and in neither has the last word been spoken.
There has been a propensity for heterogeneous units within which groups are territorially concentrated to be somewhat more inclined to fission than are homogeneous units. In India, there has been the Andhra movement in Madras, the linguistic states movement in general, the division of Punjab into Haryana and a truncated Punjab, and the fragmentation of Assam into a half dozen units along ethnic lines. Strong
demands for new states in Nigeria also emanated from sharply heterogeneous states like Kaduna, Cross River, Rivers, and Bendel. But more homogeneous states have not been immune from fissiparous tendencies_witness, in India, the Saurashtra movement in Gujarat and the
Mysore movement in Karnataka, or, in Nigeria, the Enugu movement in Anambra and the Oshun movement in Oyo. Absent preexisting states, principalities, emirates, or sultanates, to whose traditional boundaries the new units may cleave, there is a tendency for ethnic groups and prominent subethnic groups to advance claims to statehood that, if granted, might result in a greater-than-anticipated number of largely homogeneous states. Very often, however, such claims can be resisted or placated with much less than separate statehood: strong separatist movements are common but not universal.29
Within limits, the more states there are, the greater will be the tendency of ethnic and subethnic groups to be concerned with parochial alignments and issues, and the greater will be their difficulty of combining across state lines to make coherent and divisive claims at the center. All else being equal, therefore, it is probably better to have more rather than fewer states.30 Yet there is wisdom in Sir Geoffrey Vickers' observation that, in reorganizing institutions, "it is easiest to subdivide, more difficult to combine and most difficult to carve up and regroup the constituents of a going concern."31 It may be desirable to end up with a large number of units but prudent to begin with fewer.
The Costs of Federalism
In most Asian and African states, there are so many obstacles to decentralization that one hardly needs to call attention to the costs of federalism here. Yet there are costs: duplication of function, expenses of build-
29. For an example of placation, see Howard Spodek, " 'Injustice to Saurashtra': A Case Study of Regional Tensions and Harmonies in India," Asian Survey 12 (May 1972):
416-28. , ,  , ■      • ,
30. There is a related question regarding when to encapsulate a whole group in a single state and when to split it between two or more states. There may be a tradeoff here between the need to break up a large group and the desire to quarantine conflict within state boundaries. The former is aptly illustrated by the allocation of the Hausa, Yoruba, and Ibo to more than one state each. The latter is illustrated by Indian federalism. In those Indian states that encapsulate whole groups, a conflict that flares up typically does not spill across state boundaries. Not all groups are so encapsulated, however. There are Bengalis in Assam, as well as in West Bengal, and anti-Bengali agitation and violence in Assam cannot be confined to that state. Likewise, tensions between "backward castes" and upper castes in North India tend to cross state lines, particularly in Uttar Pradesh and Bihar, so that conflicts in one state reverberate in the other.
31. Vickers, The Art of Judgment (London: Chapman & Hall, 1965), 59.
622
Ethnic Groups in Conflict
ing state capitals, various diseconomies of small scale. There is also, however, an ethnic conflict cost that should be underscored.
One of the strongest forces for devolution is the expectation that government offices in lower-level units will be composed differently from central bureaucracies. The assumption may simply be that, if nature takes its course, the composition of the state civil service will generally resemble the ethnic composition of the region or state. To make this happen, no discrimination is necessary—hence the appeal of devolution for achieving ethnically distributive goats—although a language of administration different from the central official language (Tamil in the Tamil districts of Sri Lanka, for example) may be a facilitating condition. In practice, however, ethnic discrimination may occur, affecting particularly the fortunes of well-educated groups willing to migrate in search of opportunities outside their own states. This has been a serious problem in Nigeria. Some states had hired out-of-state employees on limited-term contracts, rather than on the terms available to domiciliaries of the state. In the 1970s, the military government and then the new constitution put an end to discriminatory terms of employment, but there is no guarantee that applicants from out of state will be hired at all. The other side of this particular Nigerian coin was a growing parochialism, signified by an aphorism enjoining civil servants to serve in their own states: "Don't be a Peace Corps."32
Another form of discrimination is perhaps more likely where regional autonomy schemes are implemented. An assumption may grow that members of groups with "their own" regions have no claim to work in the central government or anywhere outside their region. Such a reaction was certainly feared by those Ceylon Tamils who consistently opposed federalism. Ethnic discrimination and its attendant inefficiencies are thus facilitated by territorial boundaries.
I shall deal more extensively with the costs and benefits of distributive policies later. Suffice it to say here that, to the extent that territorial arrangements have an ethnically distributive impact, they are not exempt from these costs and benefits.
DEVOLUTION TO AVERT SEPARATISM
Important as it is to ask how a regime of devolved power should be structured, it is also important to ask related questions: when and how
32. lam grateful to Dr. Martin Dent for a letter, |an. 6, 1978, containing the quotation and a discussion of the problem and to Dr. Anthony Oyewole for a helpful conversation in Ife.Jan. 16, 1978.
Structural Techniques to Reduce Conflict
623
devolution and other policies to counter separatism can be put in place and how they can avoid fostering the very secession they aim to prevent. Proposals for devolution abound, but more often than not devolution agreements are difficult to reach and, once reached, soon abort. Most such agreements are concluded against a background of secessionist warfare or terrorist violence. Where central authority is secure, as in India, the appropriate decisions can be made and implemented by the center. But, where the very question is how far the writ of the center will run, devolution is a matter of bilateral agreement, and an enduring agreement is an elusive thing.
The recent history of Arab-Kurdish relations in Iraq is laced with failed agreements for Kurdish autonomy—agreements unimplemented by Baghdad or later rejected by the Kurds when their prospects in warfare seemed better. The federation of Eritrea with Ethiopia was resisted from the first by Muslim Eritreans and was later undone by Addis Ababa, which forced Eritrean integration into the Empire. In the course of the Moro rebellion, the Philippine government put into effect a regional autonomy scheme; it was quickly rejected as inadequate by the Moro National Liberation Front that had sought it. Only slowly have some MNLF factions begun to accept it. Bad timing, reluctance to relinquish critical areas of central control, or the prospect of additional external assistance for the separatists can easily undo such arrangements.
It takes some special conditions to create a federal or regional autonomy arrangement that will take hold. A more general constitutional change provides an auspicious setting in which to consider new territorial arrangements to cope with ethnic problems. New states were created with alacrity as the Nigerian military began the process of turning power back to civilians. The democratization of Spain was a suitable occasion to provide for regional autonomy for the Basques, Catalans, and Gali-cians. In Sri Lanka, adoption of a new presidential constitution and the increasing role of district ministers comprised the background against which power could be confided to district-level development councils. A long period of warfare that brings home the destructive impact of unrestrained ethnic conflict can sometimes serve to catalyze change (Nigeria and the Sudan), but not always (Burma and Iraq). Where separatist warfare is ongoing, a decline in the capacities of the antagonists may be conducive to a regional autonomy settlement. The Sudanese regional autonomy agreement was concluded at a time when Uganda had limited Anyanya access to supplies. The Numeiry regime in Khartoum had barely survived a Communist coup attempt and had to contend with a
624
Ethnic Groups in Conflict
continuing threat from the Ansar sect. Both sides therefore felt weaker than they previously had, and regional autonomy seemed more attractive than pursuing each side's preferred alternative, with its high costs and uncertain outcome.33 Policymakers can search for opportunities, but it is difficult to create them out of whole cloth.
The problem is complicated, as the Sudanese example well shows, by the presence of two sides that must agree. Different techniques are applicable to securing the acquiescence of ethnic groups influential in the central government, on the one hand, and gaining the acceptance of the separatists, on the other.
Regional autonomy or federal arrangements are often viewed as undue concessions to separatist sentiment. They may entail a diminution of sovereignty, or confer what seem like special privileges on troublesome and disliked ethnic groups, or "strengthen centrifugal forces and play into the hands of the separatists."34 The widespread fear that regionalism or statehood will merely feed the secession is difficult to dispel, but there are partial answers. One is for the central government to retain ultimate control over the powers of regional governments, as central governments were able to do in the Sudan, Sri Lanka, and Spain, without losing the cooperation of the beneficiary groups. Another way to reduce opposition to regional autonomy is to make it available not only to separatist regions but to all regions. The Sri Lankan development councils were to be operative throughout the country, and Spanish regional government has been offered to all regions on a referendum basis. The Sudanese scheme followed a local government law decentralizing authority to all the provinces.35 Here, however, there is a tradeoff, exempli-
33. Numeiry had executed his Minister of Southern Affairs, who was a Communist, replacing him with another Southerner (Abel Alier) who came to enjoy his confidence. Even before the agreement, Numeiry signaled willingness to provide funds for the South and had enacted a Local Government Law that decentralized authority to the provincial level. On some difficult issues, such as language, the regional autonomy agreement was ambiguous but conciliatory to the South; and it made the regional council executive subject to Numeiry's authority. But the agreement also offered the separatist guerrillas the chance to join the Sudanese armed forces. See generally Sulton, "Regional Autonomy in the Southern Sudan"; Kasfir, "Southern Sudanese Politics Since the Addis Ababa Agreement"; John Howe)), "Politics in the Southern Sudan," African Affairs 72 (Apr. 1973): 163-78.
34. Astri Suhrke, "The Muslims in Southern Thailand: An Analysis of Political Developments, 1968-78" (unpublished paper, Washington, D.C., Dec. 1978), 4.
35. Subsequently, the Sudanese devolution ran into difficulty partly because of asymmetry between the Northern and Southern regions. The Self-Government Act of 1972 created one region in the South. Five were created in the North. In the early 1980s, a move to divide the South into several regions, comparable to the Northern regions, created considerable apprehension in the South. See Dunstan Wai, "Geoethnicity and the Margin of Autonomy in the Sudan," in Donald Rothchtld and Victor A. Olorunsola, eds., Slate Versus Ethnic Claims- African Policy Dilemmas (Boulder: Westview, 1983), 304-30.
Structural Techniques to Reduce Conflict 625
fied by earlier Sri Lankan attempts to regionalize and resistance to re-gionalization in Spain. By universalizing regional autonomy, it is possible to dispel ethnic opposition to special concessions, only to solidify opposition based on the generalized fear of politicians that rival power centers are being created or that central authority will be dissipated altogether. Opposition based on either ground can doom the scheme.
The Sri Lanka scheme might not have been possible without changes in the structure of the central government. Previous attempts to devolve power on Tamil authorities had failed because Sinhalese opposition had produced back-bench revolts. Since 1978, however, Sri Lanka has had a presidential system and a constitutional provision that prevents legislators, elected on a party list system, from crossing the aisle without risking their seats.36 As intended, the result is much greater latitude for "unpopular decisions"37 between elections. The prospects for policies to reduce conflict are closely related to the character of broader structural arrangements.
Precautions such as a fixed presidential term may make it possible to offer regional autonomy or federalism, but they are no guarantee of its acceptance by separatists. Here some distinctions, based on timing and the character of the secession, are in order. An early, generous offer of autonomy, made before extreme separatist organizations outflank moderate leaders, may avert secession.38 A similar offer, made after separatist violence has broken out, may well do what opponents of concessions fear: it may testify to the weakness or vacillation of the central government and the success of the separatists, thereby fortifying their will to
36. The Constitution of the Democratic Socialist Republic of Sri Lanka, section 99 (13) (Colombo: Department of Government Printing, 1978), as amended by the Second Amendment to the Constitution, Feb. 22, 1979.
37. "To ensure the stability of the Government between elections was very important because in all developing countries Governments had to take what may be termed unpopular decisions . . . ." Statement of H.E. Mr. J. R. Jayewardene, President of Sri Lanka, at the Commonwealth Heads of Government Regional Meeting, Sept. 1980 (mimeo., n.d.), 2.
38. For the succession of increasingly intransigent organizations, each outflanking the one that preceded it, see, e.g., Brian Crozier, The Rebels: A Study of Post-War Insurrections (Boston: Beacon Press, 1960), 85-89; Martin R. Doornbos, "Protest Movements in Western Uganda: Some Parallels and Contrasts," in Raymond L. Hall, ed., Ethnic Autonomy—Comparative Dynamics: The Americas, Europe and the Developing World (New York: Pergamon Press, 1979), 274. Somewhat the same thing can be said for a policy of repression. Early, decisive, and consistent use of force against separatism seems to deter separatist warfare; late, equivocal, or sporadic repression (alternating with concession) seems merely to strengthen the separatists' resolve, as it surely did in Burma and the Sudan. Ct. Joane Nagel, "The Conditions of Ethnic Separatism: The Kurds in Turkey, Iran, and Iraq," Ethnicity 7 (Sept. 1980): 279-97, esp. 295-96.
626
Ethnic Groups in Conflict
Structural Techniques to Reduce Conflict
627
fight on. Years after warfare broke out in Chad, the Tombalbaye regime devolved revenue and judicial powers to the North, without any effect on separatism.39 In addition, there is a greater chance that regional autonomy will reduce separatist sentiment among late, reluctant secessionists like the Ibo and the Ceylon Tamils than among early, eager secessionists like the Chadian Muslims, the Karens, and the Moros.40 Policy latitude contracts much more quickly with early secessionists.
Federalism or regionalism will be most attractive if it is coupled with policies whose effect is to raise the costs of a successful secession. Here it is possible to learn something pertinent to policy from negative cases. Why is it that the Luo in Kenya, whose home is in the West and who resent the dominance of the Kikuyu, have never "seriously contemplated a Biafra-type secession"?41 Clearly, it is because they hold influential positions in major Kenyan towns outside their region, especially Nairobi and Mombasa. Like the Lozi of Zambia, but unlike the Ibo, ethnic conflict has not forced them to return home. Secession is less attractive if it is likely to mean a forfeiture of abundant opportunities outside the home region.
This lesson has not been lost on some regimes seeking to counter separatism. Pakistan has continued the British practice of providing Pa-thans with opportunities in the army, frontier scouts, and militia, while at the same time expending disproportionate funds on investment in Pathan areas, so that other groups even speak sarcastically of the government as "Pathan Raj."42 The Nagas have had similar treatment from the Indian government: a package of statehood, investment, and reserved offices outside of Nagaland.43 Once again, structural and distributive policies are used in tandem.
In point of fact, the two principal disincentives to secession are dispersion of the separatist group's population outside of the separatist region, especially in lucrative opportunities, and the regional investments or subsidies that a separatist region would lose if it opted out. It is revealing that Kurdish demands in Iraq recurrently embody claims to increased
39. For a discussion of these matters, I am indebted to a conversation with a Chadian economist in Washington, D.C., July 19,1971.
40. For the basic distinction between these two types of separatist groups, see Chapter 6, above.
41. David Parkin, "Congregational and Interpersonal Ideologies in Political Ethnicity," in Abner Cohen, ed., Urban Ethnicity (London: Tavistock, 1974), ] 42.
42. Hugh Tinker, India and Pakistan: A Political Analysts, rev. ed. (New York - Praeger 1968), 146.
43. Far Eastern Economic Review, May 11, 1979, pp. 27-29. See also the report of Walter Schwarz, Washington Post, Mar. 3, 1974.
opportunities in Arab areas and to augmented expenditures in Kurdish areas, both in addition to autonomy.44 Conversely, the 1980 "sovereignty-association" referendum in Quebec was defeated after a campaign that emphasized the post-separatist isolation that Francophones would suffer in Ontario, New Brunswick, and Manitoba, and that made telling points about tariff protection, federal subsidies, and the ratio of funds expended in Quebec to taxes collected there.45 Even then, as a separatist leader argued, sovereignty was also a matter of "pride,"46 and the referendum outcome was not foreordained.
Some secession-inhibiting policies are easy to carry out. For advanced, population-exporting groups, this is largely a matter of limiting discrimination and preventing violence against them outside their home region, so as to protect their diaspora. For advanced regions (such as the Basque country) that complain that they are subsidizing the rest of the country, the power confided to regional authorities probably needs to include an ample measure of freedom to tax and spend. For many regions, especially the more developed regions, financial policies in general may be used to create interregional entanglements and interdependence without explicit ethnic provisions. Beyond this, the problem becomes delicate, for policy may involve ethnically preferential expenditures that can simultaneously do too little to prevent separatism and yet so much as to provoke an ethnic reaction to the policies. (Here, again, we are up against the costs and benefits of preferential policies, a subject 1 shall deal with in Chapter 16.)
Territory constitutes a framework in which incentives and disincentives operate. These do not always determine group decisions to secede.
44. The Kurdish demand is usually for proportional representation in government positions and proportional per capita expenditure in Kurdish areas. Lorenzo Kent Kimball, The Changing Pattern of Political Power in Iraq, 1958 to 1971 (New York: Robert Speller, 1972), 141-42; Charles M. Benjamin, "The Kurdish Non-State Nation" (unpublished paper presented at the 1975 annual meeting of the International Studies Association), 6; Abdul H. Raoof, "Kurdish Ethnic Nationalism and Political Development in Republican Iraq" (unpublished paper presented at the 1971 annual meeting of the Middle East Studies Association), 4,10.
45. It was said that, if Quebec seceded, the Franco-Ontarians would become homeless "Palestinians of North America." The Gazette (Montreal), Apr. 22,1980. To allay fears of Francophones in New Brunswick, the Parti Quebecois (PQ) announced its support of a separate Acadian state there. Ibid., Feb. 1, 1980. Quebec's dependence on the Ontario market, on federal oil and health subsidies and pension supplements, and on spending in Quebec above Quebec's tax contributions were all stressed in the anti-separatist campaign. The PQ conceded the last point, but argued that the relevant measure was lower federal spending per capita compared to that in other provinces, rather than expenditures against revenues. Financial Post (Toronto), Mar. 8, 1980.1 am indebted to Barbara G. Haskel for an array of press clippings on Quebec ethnic issues.
46. Jacques Parizeau, quoted in The Gazette, Jan. 22, 1980.
Ethnic Groups in Conflict
628
Moreover, different sorts of incentives and disincentives are apt for different kinds of separatist groups. Policy measures will not always work or may not work when policymakers are ready to have them work. Nonetheless, the broader point remains: the most potent way to assure that federalism or regional autonomy will not become just a step to secession is to reinforce those specific interests that groups have in the undivided state.
ELECTORAL SYSTEMS AND CONFLICT REDUCTION
Electoral systems have a role in fostering or retarding ethnic conflict. The delimitation of constituencies, the electoral principle (such as proportional representation or first-past-the-post), the number of members per constituency, and the structure of the ballot all have a potential impact on ethnic alignments, ethnic electoral appeals, multiethnic coalitions, the growth of extremist parties, and policy outcomes.
Unfortunately, the development of this fertile field has been arrested by two fortuitous circumstances. First, in Asia and Africa, concern for minority electoral protection initially took the form of imposition of a separate electoral roll for Muslims in India and later for certain minorities elsewhere.47 This began a long and sterile debate over whether the so-called communal roll widens or narrows existing gaps between ethnic groups.48 Second, scholarly studies of electoral systems have had a completely different focus: whether proportional representation is more conducive to the proliferation of parties than is the first-past-the-post system
47. In 1906, the British in India were first confronted with a demand for a separate Muslim electoral roll, the first of many such demands they encountered in the colonies. See Sir Reginald Coupland, The Indian Problem, vol. 1 (London: Oxford Univ. Press, 1942), 28-36. Although they yielded in India, the British eventually resisted in other colonies, such as Ceylon, which had communal rolls until the Donoughmore Commission, in drafting the universal suffrage constitution adopted in 1931, set its face decisively against them. Ceylon Tamils accordingly boycotted the first elections under the constitution. Robert N. Kearney, The Politics of Ceylon (Sri Lanka) (Ithaca: Cornell Univ. Press, 1973), 28-33. As independence approached, the British tended to accede to separate rolls or seats only where white settlers insisted, as in Kenya, or where the British saw self-government as impossible without them, notably in Fiji and Cyprus. See Glen Wright, "Fiji Approaches Independence," Race Today (London), Apr. 1970, pp. 114-15; "Fijian Independence," Current Notes on International Affairs (Canberra) 41 (Sept. 1970): 461-64; Adamantia Pollis, "Intergroup Conflict and British Colonial Policy: The Case of Cyprus," Comparative Politics 5 (July 1973): 575-99.
48. The debate was sterile because it was abstruse and acontextual. A splendid example of the debate is W. J. M. Mackenzie, "Representation in Plural Societies," Political Studies 2 (Feb. 1954): 54-69.
Structural Techniques to Reduce Conflict
629
I
1
of election.49 Although this debate has some relevance to the impact of electoral formulae on ethnicity, it has scarcely touched explicitly on ethnic variables at all.50
Politicians have been more acutely aware of the ethnic impact of electoral provisions. Various forms of electoral manipulation and gerrymandering have been practiced in many countries to favor one group or another. More recently, electoral innovation has been used by a small but growing number of severely divided societies in Asia and Africa as a vehicle for ethnic accommodation.
There are several possible goals of such innovations, and there has not always been great clarity about them. All of the goals stem from the growth of ethnically based parties in severely divided societies.
Suppose, once again, that two groups, A and 6, support their respective parties, A and B. Suppose further that Group A comprises 60 percent of the population and a majority of voters in 60 percent of the single-member electoral constituencies, with members elected on a first-past-the-post formula; Group 6 comprises a 40 percent minority overall and a majority in 40 percent of the constituencies. Clearly, as the election returns come in, Group A and its ethnic party appear to have gained power for the indefinite future. This, as we have seen, is a simplified version of the situation that prevailed in many Asian and African countries under free elections after independence. Ethnic parties developed, majorities took power, and minorities took shelter. It was a fearful situation, in which the prospect of minority exclusion from government, underpinned by ethnic voting, was potentially permanent. Variants of this situation were responsible for much of the instability in the post-colonial world in the first ten years of independence. Civil violence, military coups, and the advent of single-party regimes can all be traced to this problem of inclusion-exclusion. Now the question is whether anything can be done about it while free elections prevail.
In principle, there are three solutions to the pure form of the problem depicted by the 60—40 split. The first is an alternation scheme, such as
49. See Maurice Duverger, LTnfluence des systemes electoraux sur la vie politique (Paris: Armand Colin, 1954); Maurice Duverger, Political Parties, trans. Barbara North and Robert North (New York: John Wiley, 1954); Douglas W. Rae, The Political Consequences of Electoral Laws, rev. ed. (New Haven: Yale Univ. Press, 1971).
50. An outstanding exception is J. A. Laponce, "The Protection of Minorities by the Electoral System," Western Political Quarterly 10 (June 1957): 318-39. See also J. A. Laponce, The Protection of Minorities (Berkeley and Los Angeles: Univ. of California Press, 1960).
630 Erhnic Groups in Conflict
the rotating presidency that was attempted unsuccessfully among three groups in Benin in the 1960s and somewhat more successfully after civil war in Colombia. Where ethnic divisions are deep, it is unlikely that such an arrangement will prove enduring. The second possibility is an all-embracing "national government." This is what many single-party regimes said they were creating when they dissolved the opposition, but what they were actually doing was something quite different. In the absence of an external emergency of the sort that produced the wartime national government in Britain, such a contrivance is no solution for ethnically divided Asian and African countries. The third solution is to use the electoral system to encourage party fragmentation with a view to producing one of two outcomes: (1) a split in Party A, resulting in two parties for Group A, neither of them with a majority of seats—hence the need for an interethnic coalition; (2) splits in both Party A and Party B, resulting in much more fluidity and, depending on how seats are apportioned, perhaps as many as four possible interethnic coalitions. It should be noted, however, that both of these solutions depend on the capacity of the electoral system to induce splits in at least one of the parties (A). Party A, however, has the most to lose if it does split, and it is thus most likely to try to avoid splitting.
There are, however, more complex versions of the ascriptive majority-minority problem. For all of them, let us assume, as before, that ethnic voting prevails; that each group has a majority in a number of parliamentary constituencies that is proportionate to ethnic shares of the population; and that the electoral formula is first-past-the-post in single-member constituencies.
Suppose Group A is 80 percent of the population, and Group B is 20 percent, but Group A is divided into two parties competing for the A vote. This is a dangerous situation, in which ethnic outbidding can occur, but it is not necessarily solved by splitting the support of Group A, which is already split. More useful would be incentives for the parties of Group A to behave moderately toward Group B or—and the two may go together—a device that would make the votes of Group B essential for the formation of a government. An even balance between the two parties of Group A may be preferable for this purpose to a system that splits Group A's support further, perhaps enabling the strongest of the A parties to win a majority of seats by recurrent pluralities in three- or four-way contests.
Another variant of the ascriptive majority-minority problem, a corn-
Structural Techniques to Reduce Conflict 631
mon one in fact, entails a situation in which several ethnic groups, none a majority, are potential contenders for power. There are, however, affinities among Groups A, B, and C, on the one hand, and D, £, and F, on the other. Given the exigencies of forming governments under the parliamentary system, it is possible that the two clusters will form two ethnically based parties, thus producing a situation as polarized as the 60-40 situation. Two courses of prevention suggest themselves. One is to create incentives for multiethnic support that will cross the chasm between the two clusters, making consolidation into two parties less likely. The other is to make it less exigent for the existing groups to congeal into a majority. Perhaps there is an electoral formula that can help preserve a more fluid multigroup-multiparty system.
So far, for simplicity, 1 have hypothesized situations in which ethnic group percentages get translated into the same fractional shares of votes for the party or parties of each ethnic group, which then get translated into the same fractional shares of seats. Thus, a group with 60 percent of the population was hypothesized, first, to cast 60 percent of the total vote for its party, and that party was assumed, second, to win 60 percent of the seats. In point of fact, there is distortion at both interchanges. Demographic and behavioral differences account for the first distortion; the electoral system typically introduces the second.
Even in conditions of acute ethnic conflict, with ethnic parties, ethnic group percentages of a population do not convert perfectly into percentages of a vote. Three variables intervene: (1) relative shares of eligible voters, which, holding registration rates constant, is essentially a function of the age structure of each ethnic group (groups with the largest percentage of members under the age of eligibility obviously have a smaller share of voters than of population); (2) relative rates of voter turnout, which vary with party organization, urban or rural concentration, and certain cultural features (in some groups, for instance, it is more difficult to induce women to vote); and (3) relative rates of voting for ethnic parties (even in high-conflict cases, the incidence of ethnic voting varies marginally from group to group). There is not much an electoral system can do about these variables, which are typically not decisive in polarizing elections in any case, and I shall not consider them further. For the sake of the discussion, 1 shall simply assume no difference between shares of a population and shares of a vote.
The translation of votes into seats is a wholly different matter, and it gives rise to yet another majority-minority problem. Quite commonly, a
632 Ethnic Groups in Confli
party with a mere plurality of votes obtains a solid majority of seats.51 Suppose, then, in a 60—40 situation, the support of Group A is divided equally between two parties, while Group B's support is given solidly to one party. In three-sided contests in single-member constituencies, under a first-past-the-post formula, Group B's party, with 40 percent of the vote, may end up with as much as 60 percent of the seats. So a cohesive minority group can govern a less cohesive majority group, or a group with a plurality of the vote can gain a majority of seats if the balance of the vote is at all fragmented. Both of these outcomes are generally perceived as instances of illegitimate minority-group rule. Where they occur, they tend to create explosive situations. Several very serious ethnic riots are traceable to electoral results of this kind. The question is whether the electoral system can be modified so that shares of votes are translated more faithfully into shares of seats.
These hypothetical examples give rise to essentially five possible aims for an electoral system that is to be harnessed to the goal of ethnic accommodation. An attempt can be made to utilize the electoral system to:
1. Fragment the support of one or more ethnic groups, especially a majority group, to prevent it from achieving permanent domination.
2. Induce an ethnic group, especially a majority, to behave moderately toward another group and engage in interethnic bargaining.
3. Encourage the formation of multiethnic coalitions.
4. Preserve a measure of fluidity or multipolar balance among several groups to prevent bifurcation and the permanent exclusion of the resulting minority.
5. Reduce the disparity between votes won and seats won, so as to reduce the possibility that a minority or plurality ethnic group can, by itself, gain a majority of seats.
The available evidence indicates that many of these aims can probably be achieved. That the aims are vitally important cannot be gainsaid. The alternative frequently is for ethnic party systems of the exclusionary 60— 40 sort to produce violent reactions and narrowly based military or
51. Rae, The Political Consequences of Electoral Laws, 75-76. In Rae's sample of Western countries, every party with more than 48 percent of the vote secured more than 50 percent of the seats. Sometimes a party with as little as 40 percent of the vote was able to gain more than 50 percent of the seats. Occasionally, a party could gain as much as 73 percent oi the seats with less than half the votes. Ibid.
Structural Techniques to Reduce Conflict 6.33
single-party regimes. I intend to consider the feasibility of the five aims through an examination of electoral arrangements in four severely divided societies in four different regions of the world: Lebanon, Nigeria, Sri Lanka, and Guyana.
FLUIDITY THROUGH FIXITY: LEBANON
From the National Pact of 1943 to the civil war that began in 1975-76, Lebanon had an electoral system that encouraged moderation, that practically required interethnic coalitions, and that prevented the crystallization of allegiances around the overarching affiliations of Muslim versus Christian. Four electoral provisions were conducive to these results: reserved offices, reserved seats, interethnic tickets, and interethnic voting.52 All the major offices were reserved. The president was to be a Maronite, the prime minister a Sunni, the speaker of the house a Shiite, the vice-speaker a Greek Orthodox, and so on. By the same token, the ethnic composition of the legislature was prescribed by law; that of the cabinet, by custom. Although there were variations over time, generally most constituencies were multimember and multiethnic. The ethnic identity of each seat was specified. There was a common electoral roll, so that each voter, regardless of ethnic identity, cast a ballot for each seat. Candidates formed competing interethnic lists, appealing to the entire electorate.
Rarely in a severely divided society has there been a system that placed as high a premium on intraethnic competition and interethnic cooperation. That the major political offices were reserved for members of specified groups meant that it was not possible for members of other groups to aspire to the same office and use ethnic appeals to mobilize support. Sunni aspirants for the prime ministership sought to link up with Maronite aspirants for the presidency, each trading the support he could muster
52. For the structutes of Lebanese politics, see Michael C. Hudson, The Precarious Republic: Political Modernization in Lebanon (New York: Random House, 1968); Leonard Binder, ed., Politics in Lebanon (New York: John Wiley, 1966); David R. Smock and Audrey C. Smock, The Politics of Pluralism: A Comparative Study of Lebanon and Ghana (New York: Elsevier, 1975); Ralph E. Crow, "Religious Sectarianism in the Lebanese Political System," Journal of Politics 24 (Aug. 1963): 489-520; Michael W. Suleiman, "The Role of Political Parties in a Confessional Democracy: The Lebanese Case," Western Political Quarterly 20 (Sept. 1967): 682-93; Michael C. Hudson, "Democracy and Social Mobilization in Lebanese Politics," Comparative Politics 1 (Jan. 1969): 245-63; Enver M. Koury, Crisis in the Lebane.se System: Confessionalism and Chaos (Washington, D.C.: American Enterprise Institute, 1976); Riad B. Tabbarah, "Background to the Lebanese Conflict," International Journal of Comparative Sociology 20 (Mar.—June 1979): 101—21.
634 Ethnic Groups in Conflict
in his own group for the support of the other. Reserved offices reinforced tensions within each group, while solidifying ties between one group and another. The same was true of legislative elections.. Reserved seats restricted competition to members of the same group. Mixed tickets in multimember constituencies with a common roll required agreements among politicians of various groups—underpinned, of course, by the need for every candidate to obtain votes of members of several groups. The fixed proportions of the system meant that it was impossible to increase the number of seats held by any group or for a group to occupy an office or seat assigned to another group. All that was left was to squabble over who, among members of a given group, would occupy a seat or office and, in the process of doing so, to maximize support from sources outside the group.
This was a system that depended upon and exacerbated preexisting subethnic cleavages, based on family, clan, and region. Lebanon is richly endowed with such distinctions, and so it was by no means artificial to force one Maronite faction to oppose another, for instance, or to induce the Chouf Druze to line up against their Druze rivals, the Yazbakis. Lebanese subgroups were tailor-made for this system, and they rose to its opportunities.
With half a dozen major groups and many subgroup factions, typically organized around strong leaders, there was much flux in the system. The cabinet was, of course, composed of fixed proportions of the various groups. There were many candidates competing for inclusion in it, and so it was possible for the president to maintain his power by reshuffling the cabinet frequently. This flux in appointments was a disincentive to extreme opposition. A Shiite, for example, excluded from the cabinet today might be required for a Shiite position in it tomorrow. Extreme behavior could only impair his future attractiveness. Consequently, the same system that made subethnic factions the most relevant political actors encouraged rotation in office and discouraged ethnic appeals that might alienate potential electoral, legislative, or cabinet allies.
Similar forces inhibited the growth of political parties. It was the individual leaders who, by dint of ethnic and subethnic identity, were asked to fill reserved places. This enhanced the position of notables who commanded factions. Occasionally, they called their faction by a party name. (Some fifteen to twenty "parties" could be identified in the 1960s.) But party organization would have deprived such leaders of the flexibility they needed to make and unmake electoral arrangements and to enter
Structural Techniques to Reduce Conflict 635
and leave cabinets as individuals.53 Moreover, there was no acceptable ethnic basis for party organization. Since ethnic hostility was strong, it was not possible to create parties that were multiethnic in support. Yet, because of the constant need to forge links and secure votes across ethnic lines, neither was it possible to create ethnically based parties. The system had to be personalistic.
The Lebanese system provided a deeply divided society with a politics of conciliation for over thirty years.54 It was undone by many forces, external and internal, most of them unrelated to the character of the electoral system.55 Two, however, bear mention. Neither is inherent in a Lebanese type of system.
First, the absence of parties created an organizational vacuum, which facilitated the emergence of armed private militias. These gangs, some of them attached to parliamentary politicians, had much greater freedom than they would have had if real party organizations had existed and seen them early on as rivals for political authority. The notorious weakness of the Lebanese state, related to its ethnic structure, was also conducive to the flourishing of militias.
Second, as indicated in Chapter 14, the fixed ethnic proportions of the system were a given. Small adjustments in parliamentary representation were made from time to time, but the overall proportional shape of the system, with its reserved offices, could only be challenged by challenging the entire structure. In a sense, Lebanese electoral institutions created two options: moderation or civil war. The former was much more common, but the latter, when it came, was deadly.
Iv^ULTIETHNICITY THROUGH DISTRIBUTION: NIGERIA
A new constitution went into effect in Nigeria in 1979. It was abrogated by military intervention less than five years later. In the interval, however, a major electoral experiment went forward.
Determined to avoid a recurrence of the ethnically exclusionary politics that had produced the Biafra war, the framers of the new constitution attempted a far-reaching reconstruction of the political system. Nigeria's parliamentary system was replaced by American-style institu-
53. Hudson, The Precarious Republic, 148, 265-66.
54. The brief 1958 civil war was an exception, but it was largely brought on by a breach of the rules by the then-president.
55. See Richard Hrair
IDekmejian, "Consociational Democracy in Crisis: The Case o
Lebanon," Comparative Politics 10 (Jan. 1978): 251-65.
636
Ethnic Groups in Conflict
tions, replete with a separately elected president, a two-house National Assembly, and a strong separation of powers, together with comparable institutions in the nineteen states. The reasoning was that the parliamentary system had been conducive to ethnic divisions. Whereas any party with a parliamentary majority—even an ethnically limited majority— could form a government and rule the rest, a nationally elected president might be more broadly representative. Even if not, the separation of powers could prevent an ethnic group dominating one branch from controlling everything.
The framers of the Nigerian constitution aimed at a multiethnic party system. In this they did not succeed, but they did not wholly fail either. A mixed electoral system produced mixed results. And indeed the framers seem to have succeeded in something they did not quite intend: preserving the fluidity of a multiplicity of ethnic parties against the threat of a North-South bifurcation.
The principal device to encourage multiethnic parties was the presidential electoral formula.56 To be elected president, a candidate was required to win a plurality of votes nationwide plus at least 25 percent of the vote in no fewer than two-thirds of the nineteen states. Since no one or two ethnic groups (even in combination) had voters distributed widely enough to meet this stringent requirement, the expectation was that it would produce a party system with a small number of parties, perhaps just two, each with broad multiethnic support. Otherwise, it might be impossible for any presidential candidate to get elected.57
This logic was not incorrect, but it was incomplete. Although the presidential electoral formula did create incentives to multiethnicity and party consolidation, by itself it was not sufficient to produce broadly multiethnic parties or only two parties. Other features of the constitutional structure created countervailing incentives to one or both of these goals.
56. There was also a formal requirement of geographical distribution of party membership, to discourage ethnically exclusive parties, but this, predictably, was easily satisfied and was no impediment to any of the ethnically based parties. For the new Nigerian constitution, see Donald L. Horowitz, "About-Face in Africa: The Return to Civilian Rule in Nigeria," Yale Review 68 (Winter 1979): 192-206.1 am also drawing here on interviews I conducted in Nigeria in 1978 and 1980.
57. The framers were much concerned with the possibility of deadlock, and they considered a number of different presidential electoral formulae to minimize the possibility. See ibid., 197-201. In the end, they settled on a vote of federal and state legislators if no candidate won on the first ballot, but the departing military government prescribed a runoff election instead.
Structural Techniques to Reduce Conflict
637
The proliferation of states, as we have seen, made it possible for minority parties in the North to control states in the 1979 elections, thereby fragmenting party support in the North and insuring party proliferation overall. In fostering multipolarity, the new states served a function similar to the Lebanese electoral arrangements.
The two major Nigerian innovations thus pulled in opposite directions: the presidential distribution requirement toward fewer parties, the new states toward more. Overall, the incentives to party consolidation were not strong. With more than two presidential candidates, only a plurality was required to win. The candidate who did win in 1979 had only a third of the total vote, and in 1983 he still had less than half. The winner had to have broadly distributed support, but the party system did not have to congeal into two or three parties to produce a winner. Once a plurality president was installed, he might have to arrange a legislative coalition to get his program through, for his or another party might only have a plurality in the National Assembly, but his appointment power gave him bargaining leverage to negotiate a coalition; and, after the 1979 election, this was done. The coalition, rather short-lived, resembled Nigeria's earlier coalition of convenience.
The incentives to multiethnicity were also countered. There was, to be sure, no ignoring the distribution requirement for presidential election. All parties sought support outside their core region, but two did this better than the others. In the two elections held under the new constitution, 1979 and 1983, the mainly-Ibo party (NPP) drew considerable support in a Middle Belt state, Plateau, where it was also able to elect a governor. The mainly-Hausa party (NPN) developed strong links outside the Muslim North—in Rivers, Cross River, and Benue states— and came in first or second in nearly every state, in both presidential and senatorial contests. The NPN candidate, Shehu Shagari, was the only presidential contender even to approach the 25 percent distribution requirement. In 1979, he had the largest number of votes, more than 25 percent in twelve states, and about 20 percent in the thirteenth state. This, the electoral commission decided, was tantamount to having 25 percent in two-thirds of the states.58 In 1983, Shagari won at least 25 percent in sixteen states, but some of those results were doubtful.
58. Unofficial returns of the presidential election may be found in West Africa, Aug. 27, 1979, p. 1573. For an analysis, see Richard A. Joseph, "Democratization Under Military Tutelage: Crisis and Consensus in the Nigerian 1979 Elections," Comparative Politics 14 (Oct. 1981): 75-100. For the 1983 presidential results, see West Africa, Aug. 15, 1983, p. 1866.
638
Ethnic Groups in Conflict
The presidential election was not the only election, but it was the only one to have the distribution requirement. Both houses of the National Assembly were elected on a first-past-the-post formula in single-member constituencies. These constituencies generally were ethnically homogeneous or had a majority or large plurality ethnic group. A presidential candidate might take his constituency to be all of multiethnic Nigeria. A representative or senator could hardly do the same. Inevitably, most members of the legislature would arrive in Lagos as delegates of their ethnic groups and would expect their parties to be responsive to ethnic claims.
The party system that emerged from this interplay of ethnic groups and electoral incentives was mainly a somewhat realigned, expanded, and newly balanced version of the earlier ethnic party system. Indeed, it is significant that proto-parties began to emerge as divisive ethnic issues emerged in the Constituent Assembly that redrafted the constitution.59 But the new system did not seem to have the dangerous tendency of the earlier ethnic party system to polarize into two parties—one Northern, one Southern—for four parties survived through two general elections.
The electoral system of the Second Republic was subject to strong cross-pressures. Shagari's party had a core of support in the Hausa-Fulani North. Yet, if Hausa-Fulani interests had been given untrammeled sway by the president, the support his party enjoyed outside the North would have eroded. That extraregional support was more important than it was for the predecessor NPC in the First Republic. The presidential electoral formula required it. Even if some of the opposition parties had merged, the NPN's strength in the South would have made North-South bifurcation an unlikely outcome. It is not surprising that the president emerged as a conciliatory pan-ethnic figure or that the legislature was a forum for the expression of ethnic demands.60 The two electoral systems pushed those subject to them in different directions. Parties, which aimed to elect both legislators and presidents, were simultaneously pulled both ways. Electoral engineering in the Nigerian Second Republic modified but did not obliterate the ethnic basis of the party system. That was to be expected, since the engineers went to work only on the election of the president.
59. A split over a proposed Sharia court of appeals, favored by the Muslim North but opposed by the Middle Belt and South, helped draw the first party lines, but later splits increased the number of parties eventually formed.
60. On such matters, for example, as the composition of the civil and foreign service, which was occasionally debated angrily in the legislature.
639
MODERATION THROUGH ALTERNATIVE VOTES AND PROPORTIONAL REPRESENTATION: SRI LANKA
As the Nigerians were drafting a new constitution, so were the Sri Lan-kans. Both chose a presidential form of government, the Nigerian resembling the American arrangements, the Sri Lankan somewhat resembling the French, with a prime minister as well as a president. Both constitutions attempted in different ways to make provision for reducing ethnic tensions. Like the Nigerian constitution of 1978, the Sri Lankan constitution of 1978 adopted a different formula for electing the president from that employed in electing the legislature. But, while the Nigerian presidential formula was devised for ethnic engineering purposes, the Sri Lankan formulae responded to somewhat different concerns. Neither the presidential nor the legislative electoral provisions adopted in Sri Lanka aimed at multiethnic parties. Nevertheless, both Sri Lankan formulae are likely to have important effects on ethnic accommodation.61
Whereas the Nigerians constructed a presidential electoral formula that, they anticipated, would produce party consolidation, the Sri Lan-kans took their multiplicity of parties as a given. With the few parties they envisioned and the stringent distributive formula they enacted, the Nigerians were content to elect a president on a plurality basis, rather than resort to the additional complexities required to construct a majority if there were more than two candidates. With at least several parties, however, plurality election of the president was less attractive to the Sri Lankans, for the plurality achieved by a candidate might be a small fraction of the total vote. Only once in seven Sri Lankan general elections since independence had any party secured more than 50 percent of the total vote. The strongest party generally polled 30—40 percent. Accordingly, it was provided that the president must have a majority. But how to arrive at a majority if there are several candidates? For this purpose, a preferential or alternative vote system was adopted. Where there are three candidates, each voter must specify his second preference; where
61. The electoral innovations are described in C. R. de Silva, "The Constitution of the Second Republic of Sri Lanka (1978) and Its Significance," Journal of Commonwealth and Comparative Politics 17 (July 1979): 192-209; James Manor, "A New Political Order for Sri Lanka," The World Today 35 (Sept. 1979): 377-86; K. M. de Silva, "Political and Constitutional Change in Sri Lanka," The Round Table 273 (Jan. 1979): 49-57; W. A. Wiswa Warnapala, "Sri Lanka 1978: Reversal of Policies and Strategies," Asian Survey 19 (Feb. 1979): 178—87. The provisions are contained in The Constitution of the Democratic Socialist Republic of Sri Lanka, chap. XIV.
640 Ethnic Groups in Conflict
more than three candidates, his second and third preferences. In the absence of an initial majority, all but the top two candidates are eliminated. The alternative preferences of voters whose first (or second) choices are not among the top two contenders but whose second (or third) choices are among the top two are reallocated to them to compute a majority.
Under the previous parliamentary system, Tamil parties occasionally held the balance between the two major Sinhalese parties and were able to gain concessions in exchange for Tamil votes in Sinhalese-majority constituencies and Tamil support in parliament. More often than not, however, a plurality for one of the Sinhalese parties could be translated into a majority of parliamentary seats, thus obviating the need for conciliating the Tamils.62 Under the new system, however, presidential candidates cannot benefit from such a bonus. Consequently, interparty arrangements must be made in advance for second preference votes. Under ordinary conditions, this should give the Tamil United Liberation Front, which largely controls the Ceylon Tamil vote, and the Ceylon Workers' Congress, which controls most of the Indian Tamil vote, good bargaining power. If Tamil candidates run for president, they are unlikely to finish first or second. The second and third preferences of Tamil voters should thus become quite valuable commodities in political exchange.63 A Sinhalese president elected on reallocated Tamil votes has a strong disincentive to extremism. The alternative vote is said to be conducive to a politics of bargaining and moderation,64 and Sri Lanka's use of the technique will probably work in accordance with that view. Unfortunately, the evidence so far is not conclusive. Until the 1983 riots, Sri Lanka's first president, J. R. Jayewardene, was certainly more moderate on ethnic issues than his predecessors. But, in the first election under the
62. For a convenient summary of party votes and seats since 1947, see James Manor, "The Failure of Political Integration in Sri Lanka (Ceylon)," Journal of Commonwealth and Comparative Politics 17 (Mar. 1979): 23-46, at 45 n.99.
63. For similar assessments, see de Silva, "The Constitution of the Second Republic of Sri Lanka (1978) and Its Significance," 198-99; Manor, "A New Political Order for Sri Lanka," 381. A more cynical interpretation of the second preference system holds that Tamil second preferences will generally go to the United National Party, thus insuring election of its candidate. The constitution was promulgated by a UNP government. But this argument comes to the same thing, for if any other Sinhalese candidate wishes to be elected president, he will have to bid against the UNP for Tamil support. Under either interpretation, the provision encourages interethnic bargaining.
64. Palley, Constitutional Law and Minorities, 16-17. Cf. Laponce, "The Protection of Minorities by the Electoral System," 326-28.
Structural Techniques to Reduce Conflict 641
new system, vote transfers were not necessary, since Jayewardene won 52 percent of the vote.65
Soon thereafter, the future of interparty accommodation was put in jeopardy by the riots. Precipitated by a Tamil terrorist ambush of an army unit, the violence was no reflection on the new constitutional arrangements, which had not really been tested. But the aftermath of the riots opened a breach that made all interethnic political relations problematic.
The legislative electoral system adopted for the Second Sri Lankan Republic would also be expected to have a moderating effect on ethnic politics. As in the case of the president, the probable fate of Sinhalese candidates was linked to Tamil votes.
The principal purpose of the proportional representation scheme that was adopted was to prevent small swings in votes from producing large swings in number of seats. Sri Lanka has had a change of government at virtually every parliamentary election, and shares of seats have frequently been far out of line with shares of the total vote. As Douglas W. Rae has shown, proportional representation, like first-past-the-post, tends to inflate the strongest party's share of seats at the expense of the weakest, but PR does this less prominently than first-past-the-post; and PR does not generally magnify changes in party support when legislative seats are allocated, so it is particularly responsive to the oscillation problem the framers of the Sri Lanka constitution were addressing.66
The system adopted in Sri Lanka was the party list system by multimember territorial constituencies. To avert the party proliferation that was feared—some say to dilute the strength of the small Marxist parties—any party with less than one-eighth of the vote in a constituency is awarded no seats there. This is a rather high cutoff that disadvantages those small parties whose strength is not territorially concentrated.
In ethnic terms, PR should be conducive to Sinhalese moderation. PR in multimember constituencies tends to reduce the seat advantage enjoyed by territorially concentrated minorities, such as the Tamils.67 So, assuming the continuation of Tamil parliamentary participation, the Tamil United Liberation Front will generally not gain more seats under
65. For a report, see S. W. R. de A. Samarasinghe, "Sn Lanka in 1982: A Year of Elections," Asian Survey 23 (Feb. 1983): 158-64.
66. Rae, The Political Consequences of Electoral Laws, 88-92, 101.
67. Ibid., 170.
642
Ethnic Groups in Conflict
^1
this formula than under first-past-the-post.6S The conciliatory effects derive, rather, from the apportionment of Sinhalese seats, the character of the list system, and the probable nature of electoral appeals.
First of all, PR seems likely to achieve the intended effect of reducing the spread between shares of votes and seats. With Sinhalese seats more evenly divided between the two main parties, Tamil support should more often be pivotal to the formation and maintenance of parliamentary majorities than it has previously been.
In addition, though the Tamil parties do not gain seats under PR, there may well be an increase in the number of Tamil candidates. Indeed, in multimember constituencies with Tamil minorities, Tamil candidates might even appear on Sinhalese party lists. The constituency list system of PR makes it more attractive for parties to have Tamil candidates than it was under first-past-the-post.
Finally, for similar reasons, PR tends to encourage moderation in electoral appeals where minority voters in a constituency can be alienated by Sinhalese extremism. Under first-past-the-post, at least in some constituencies, minority voters could be ignored by Sinhalese parties without paying a penalty in seats. Now that every last vote counts, this is a less compelling strategy.
Once again, however, there has been no definitive test, for, after the 1982 presidential election, the government decided to prolong the life of parliament by six years, until 1989. This it did by winning a referendum required by the constitution. Consequently, there has yet to be a parliamentary election conducted under PR.
The new importance of Tamil second preferences, of Tamil votes for parliamentary candidates in the South, and of pivotal Tamil seats in parliament—all of which can be powerful influences in the competition between more evenly matched Sinhalese parties—should ultimately cement coalitions of commitment between the Tamil parties and the more conciliatory of the main Sinhalese parties. This, of course, assumes the continued vitality of two-party competition on the Sinhalese side. It also assumes that Tamil separatism does not break out into warfare and that Tamil parties are allowed to play a normal parliamentary role. These things can no longer be taken for granted in the aftermath of the 1983 riots, the continuing terrorism, and governmental and military hostility
68. For illustrations, see de Silva, "The Constitution of the Second Republic of Sri Lanka (1978) and Its Significance," 203; Robert N. Kearney, "The Political Party System in Sri Lank:'," Political Science Quarterly 98 (Spring 1983): 17-33, at 32.
Structural Techniques to Reduce Conflict
643
to the Tamils.69 The electoral innovations are not too little, but they might be too late.
ETHNIC PROPORTIONALITY THROUGH PR: GUYANA
There are times when first-past-the-post systems distort the electoral demography of ethnically divided societies. By inflating the share of seats obtained by an ethnic party with a majority of votes, first-past-the-post can reduce ethnic minority representation to below proportional levels. By the same process of inflation, the plurality share of the largest or most cohesive minority can be translated into a majority of seats, sometimes at the expense of a less cohesive majority of voters. In these circumstances, proportional representation can change the result in one of two ways. If it reduces the vote-seat disparity, PR can prevent ethnic minority rule by denying a majority of seats to a party with a mere plurality of votes. If, as frequently asserted, PR tends to proliferate parties, the resulting multiplicity of parties can deny every group a majority of seats, thus creating Lebanese-style fluidity and eliminating the problem of perpetual minority status. Both of these objectives—proportionality and fragmentation—were given a good test by the system designed for Guyana on the eve of its independence.70
To be sure, the unqualified assertion that PR tends to proliferate parties is not accurate, but a PR system can be designed, with large multimember constituencies and no minimum percentage cutoff, so as to maximize the chances of both strict proportionality and party fragmentation.71 Precisely that was done in Guyana. The British moved from first-past-the-post in single-member constituencies in the 1961 election to PR, with the whole country a single constituency and each party putting up a single list, for the 1964 election. It should be noted that PR with one nationwide list per party is inimical to fostering coalitions of commitment or alliances, for it prevents vote pooling in a way that indi-
69. See the issue "Sri Lanka: Racism and the Authoritarian State," Race and Class 26 (Summer 1984).
70. What follows draws on interviews I conducted in Guyana in 1965. For election results, see Report on the General Election of Members of the Legislative Assembly, 3961 (Georgetown: Government Printery, 1964); Report on the House of Assembly General Election, 1964 (Georgetown: Government Printery, 1965). See also Peter Simms, Trouble in Guyana (London: Allen &C Unwin, 1966); Cheddi Jagan, The West on Trial: My Eight for Guyana's Freedom (London: Michael Joseph, 1966); B. A. N. Collins, "The End of a Colony—II: British Guiana, 1965," Political Quarterly 36 (Oct. 1965): 406-16.
71. See Rae, The Political Consequences of Electoral Laws, 151-70.
644 Ethnic Groups in Conflict
vidual candidacies or constituency lists do not. But, in Guyana, that was not its purpose—fragmentation was. It was forecast at the time that PR would produce "a proliferation of splinter parties," that, "since each minor group can have its own representation, the society will find its many divisions increasing . . . ."72 This forecast proved entirely wrong. Fragmentation did not occur, but proportionality did.
The East Indian—dominated People's Progressive Party (PPP) of Cheddi Jagan had won a solid majority of seats in 1961 on a 42.6 percent plurality of the vote. East Indians were a majority of the population at the time, but their population was disproportionately under the voting age and slightly less cohesive in ethnic voting than was the Afro-Guyanese population, which voted heavily for the People's National Congress (PNC) of Forbes Burnham. A small third party, the United Force (UF), got many votes from the mixed (Eurafrican) population and the smaller groups of whites, Chinese, and Portuguese. Together the PNC and UF had outvoted the PPP, but, since their votes tended to be more urban and more concentrated than PPP votes were, these two parties were consigned to minority status in the legislature. This result led to considerable unrest in 1962—63.
With independence approaching, the British instituted a PR system. Jagan's Marxism had alarmed the Kennedy administration in Washington, which had urged the British to change the electoral system to effect a change of regime.73 In this, the shift to PR was remarkably successful, for it apportioned seats much more closely in accordance with votes and made possible a PNC-UF coalition government, as Table 12 shows.74
At the same time, it was thought that PR might provide the necessary inducements for Jagan's East Indian support to split into its component parts: especially Hindu-Muslim, but perhaps also left-wing and right-wing. Several parties did spring up to contest the 1964 elections. One of them was an Indian Muslim party, another an ideologically moderate Indian party. The two managed to obtain no seats and together only one
72. Peter Newman, British Guiana: Problems of Cohesion in an Immigrant Society (London: Oxford Univ. Press, 1964), 97.
73. See Arthur M. Schlesmger, Jr., A Thousand Days: John F. Kennedy in the White House (Boston: Houghton Mifflin, 1965), 645-49.
74. The PNC-UF coalition, though more or less multiethnic, did not really span the main fault line of Guyanese politics: the Creole-East Indian division. Consequently, it was not the sort of accommodative coalition 1 have been speaking of. It should also be noted, in connection with Table 12. that the comparison is restricted to 1961 versus 1964, because post-1964 election data in Guyana are not reliable.
m
TABLE IZ      ELECTORAL OUTCOMES IN GUYANA, I 96 I AND 1964 (by party; in percentages)____
First-Past-the-Post: 1961 Proportional Representation: 1964
Party	Votes	Seats	Seat-Vote Disparity	Votes	Seats	Seat-Vote Disparity
PPP PNC UF	42.6 41.0 16.4	57.1 31.4 11.4	14.5 -9.6 -5.0	45.8 40.5 12.4	45.3 41.5 13.2	-0.5 1.0 0.8
Total	100.0	99.9"	9.7b	98.7'	100.0	0.8b
"Total does not equal 100.0 because of rounding. "Average disparity for the three parties.
'Total does not equal 100.0 because of rounding and minor party votes.
percent of the vote. The incentives existed, but, as we saw in Chapter 8, the electorate set its face against fragmentation.
In fact, the only third party with any significant strength, the United Force, lost votes under PR: four percentage points. This, however, had nothing to do with the electoral formula. In 1961, the UF had benefited from some votes that would otherwise have gone to the PPP in constituencies where the PPP did not put up a candidate because Indian voters were too few to make it worthwhile. Under the list system of 1964, these
votes went back to the PPP.
With this insignificant exception, what is impressive about the Guyanese experience is how stable voting patterns remained despite the radical change in electoral system. This stability makes PR a feasible strategy to achieve proportionality where first-past-the-post has distorted the results, but it makes PR an unlikely vehicle to create a new fluidity. It is one thing to say that, starting from scratch, electoral incentives do or do not produce a given result. It is another to claim that they can rapidly undo a result already entrenched in the party system. Guyana is consistent with Maurice Duverger's conclusion that "on the whole P.R. maintains almost intact the structure of parties existing at the time of its appearance."75
75. Political Parties, 252.
I
646
THE PROMISE OF ELECTORAL CHANGE
These four experiments proceeded from various motivations. In Lebanon and Nigeria, ethnic accommodation was an explicit goal of electoral innovation. In Sri Lanka and Guyana, PR was invoked to reduce vote-seat disparities, though in Guyana for the special purpose of ousting and perhaps splitting the governing party. In Sri Lanka, the preferential vote was used to prevent capture of the presidency by a candidate with a low plurality. Such a prophylactic measure was perhaps less necessary in Nigeria, with its plurality plus distribution requirement. Although the purposes were various, the experiments illuminate the prospects for achieving the five accommodative aims enumerated earlier: fragmentation, moderation, coalition, fluidity, and proportionality.
1. Fragmentation of the support of a majority group to avert its permanent domination seems a difficult goal to achieve through electoral means. The Guyanese electoral system was as conducive to achieving this goal as any is likely to be; yet it had no significant effect on voting patterns. Of course, the structure of subgroup cleavages in Guyana was not propitious. East Indians were not significantly divided along Hindu-Muslim lines. Had there been Hindu and Muslim factions in Jagan's PPP, then PR might have had an impact. Even then, however, it needs to be remembered that the East Indians had powerful incentives to remain cohesive; had they split, they would have lost their majority status. As Duverger has noted, if a conflict is already bifurcated, party splits in response to PR are unlikely.76 Group structure is an important variable influencing the response to electoral incentives, and group interest is a variable bearing on group structure.
Fragmentation was achieved in Northern Nigeria, but not by the electoral system per se. Rather, the federal system, by providing state-level arenas for electoral victory and for power to be exercised, laid the groundwork for a somewhat more fragmented ethnic party system. Fragmentation continued despite later-enacted, though incomplete, electoral incentives for party consolidation. In addition to group structure and group interest, features of the territorial environment also affect the response to electoral incentives. The division of territory is probably a more reliable way of fragmenting the support of dangerously large
76. Ibid., 244.
Structural Techniques to Reduce Conflict
647
groups to achieve multipolar fluidity than is the introduction of a particular electoral formula.
2. The encouragement of moderation appears easier to achieve through the electoral system than is the restructuring of group support. Several approaches are possible. The elaborate Lebanese system of reserved offices and seats and mixed lists is certainly one way; the Nigerian geographic distribution requirement is another; and the Sri Lankan PR and preferential vote formulae seem to be yet another. Of course, the Nigerian provisions were tested only twice, and the Sri Lankan not yet at all. Still, the Nigerian provisions worked according to their conflicting logic, and Nigerian assumptions about a separately elected president in lieu of a prime minister were impressively confirmed. If a presidential system is designed so that the electoral formula encourages moderation and penalizes ethnic exclusivism, as it was in Nigeria and Sri Lanka, the potential for presidential systems to foster accommodation seems considerable.
3. Multiethnic parties are difficult to encourage in severely divided societies. Nigeria's electoral incentives were sufficiently mixed so as to reduce their value as a test of the possibilities on this score. Although at least two parties in the Second Republic—the NPN and, to a lesser extent, the NPP—were notably more multiethnic than their predecessors in the First Republic were, none of them spanned, with any degree of completeness, any two of the three largest groups. Even so, Nigerian parties were probably pushing close to the limits of their ethnic inclusiveness.
Multiethnic coalitions of ethnic parties are possible, however. The need for a majority of legislative seats may be sufficient to induce coalitions of convenience, but these, as we have seen, do not foster accommodation. More enduring arrangements—coalitions committed to ethnic accommodation as a policy goal—are more likely to be formed as a result of the dependence of each partner on popular votes commanded by the other, rather than just on an exchange of legislative seats.
There are at least three ways to encourage such party interdependence where it would not otherwise exist: preferential voting (Sri Lankan presidency), list-system PR in heterogeneous multimember constituencies (Sri Lankan parliament), and mixed lists in constituencies with a common electoral roll (Lebanon). Still another way is suggested by Malaysia: single-member constituencies, multiethnic in composition, so that coali-
648 Ethnic Groups in Conflict
tion partners can provide marginal votes for each other's candidates on a reciprocal basis. This enumeration readily shows how, with a bit of imagination, the electoral system might be used to promote multiethnic coalitions.
4.   The existing fluidity of a multipolar system of ethnic parties is easier to preserve than it is to create anew through efforts to fragment already consolidated parties. Not only the Nigerian experience, but the Lebanese experience, suggests this. Lebanon's subgroups were, to be sure, the main actors of politics, but there was also a tendency for higher levels of group identity to assume increasing importance. "In times of trouble," Michael C. Hudson remarks, "one is compelled to identify himself as a Christian or a non-Christian; in ordinary times it is often necessary for an individual to assert or exploit his membership in a particular sect, like the Sunnite (Muslim) or the Maronite (Christian)."77 In a different electoral system, these ultimate identifications, Christian and Muslim, could have formed the basis of party loyalty and of a bifurcated, conflict-prone party system. Comparable North-South loyalties did exactly that in the First Nigerian Republic. The elaborate Lebanese electoral system did a remarkable job of keeping the center of gravity of Lebanese politics at the lower levels of the groups and subgroups. There it stayed even at the peak of the civil war of 1975-84. The Nigerian plurality election of the president and the proliferation of states also reduced the impulse to consolidate, and thus to bifurcate, the party system.
Proportional representation can also help in such cases, by making party consolidation in a fragmented system less attractive. Some light is cast on this possibility by a sideways glance at Belgium. For many decades, the Belgian party system was organized along Left-Right and clerical-secular lines, each of the three major parties spanning ethnic divisions, albeit in different proportions. When Fleming-Walloon conflict rose to the forefront of Belgian politics in the 1960s, so did explicitly ethnic parties on both sides. By 1971, these two parties had between them gained about 20 percent of the vote. Under a first-past-the-post system, the ethnic parties would have been submerged. Even under PR, the three major parties were long able to prevent ethnic claims from producing splinter ethnic parties with substantial electoral strength.78 It
77. The Precarious Republic, 2].
78. Rae has shown that PR does not automatically make it easier for a new party to gain seats. The Political Consequences of Electoral Laws, 151—67.
Structural Techniques to Reduce Conflict
took a major surge in ethnic tension to do this. Once five parties, rather than three, were in existence, however, PR helped them stay afloat, just as it had earlier kept the third party (the Liberals) alive, thereby preventing the consolidation of opinion into only two parties.79 In multipolar countries like Belgium and like Nigeria from 1979 to 1983, governing is difficult, because a legislative majority is hard to find, but polarization is averted by the fluidity of relationships possible among contending forces. The electoral system cannot manufacture ethnic and subethnic divisions, but it certainly can help sustain them once they have crossed the threshold of political relevance.
5.   Certain versions of PR can reduce the disparity between votes and seats. Sometimes the disparity is not great, and it makes no difference to the outcome. Sometimes the disparity is significant and perhaps desirable. In Malaysia, it has strengthened the position of a permanent multiethnic coalition against ethnically based flank parties. In six parliamentary elections, first-past-the-post has provided the ruling National Front and its predecessor, the Alliance, a bonus of between 18 and 30 percent of seats over votes.80 First-past-the-post in this case countered centrifugal forces. But where all parties are ethnically based and a minority is seriously underrepresented because first-past-the-post provides a voting majority with a bonus in seats, or where a minority group rules because its plurality of votes translates into a majority of seats, then PR with a list system, large multimember constituencies, and low or no cutoffs for minor parties can help bring seats and votes into line.
There is a more intriguing question of ethnic accommodation lurking in the vote-seat disparity issue. Is it possible that this version of PR, imposed before a party system crystallizes, can actually prevent the emergence of ethnic parties? In the early evolution of ethnic parties, lines frequently are not clearly drawn. Ethnic voting may not approach 100 percent the first time around, but it may be high enough, in a first-past-the-post system, so that, if groups are territorially concentrated, the party that gains most of a group's votes will win all of its seats. This will make it fruitless for dissident members of the group to withhold support
79. George Armstrong Kelly, "Biculturalism and Party Systems in Belgium and Canada," Public Policy 16 (1967): 316-57, at 326.
80. In 1959, the Alliance won 71 percent of the seats with 52 percent of the votes; in 1964, 86 percent with 58 percent; in 1969, 64 percent with 49 percent; in 1978, 85 percent with 55 percent; in 1982, 86 percent with 61 percent. 1 have omitted 1974 because the National Front then included a Malay party, the PM1P, that in all other elections had been in opposition.
650
Ethnic Groups in Confl
from that party. Rates of ethnic voting will increase, and the identification of each party with an ethnic group will be complete. Under PR, however, dissident votes will count, and perhaps it will be harder to complete the alignment of parties with groups.
Although the point is well worth a test, I am inclined to think that the process by which ethnic parties emerge in severely divided societies cannot be retarded once a strong majority of an ethnic group casts its support for one party. Two exceptions come to mind, however.
The first is where support for such a party is limited by subethnic cleavages. If a prominently demarcated subgroup withholds its support from a party purporting to represent the whole group, PR may be enough to fortify this reluctance. Very likely, however, this will not make for nonethnic or multiethnic parties; it will only mean that this particular ethnic group will be represented by more than one party.
The second is not really an exception, for it involves a case where ethnic divisions, though significant, are not so powerful as to be the only divisions. There are countervailing forces for moderation. A number of Western countries fit this description; fewer ethnically divided Asian and African countries do. In Canada, the vote-seat disparity has been so significant as to foster regional and ethnic polarization, and PR has been recommended. Under these conditions, the remedy should work.81
It should surprise no one that electoral reform cannot work magic on ethnic conflict. It cannot, for example, solve the pure form of the 60—40 problem posed at the outset. No electoral device can split Group A's 60 percent, so that Group B will not be a perpetual minority. Yet it is extraordinary just how much an electoral system can do in what seems an
81. In the 1980 Canadian parliamentary elections, the victorious Liberals, with 23 percent of the vote in the Western provinces, gained only two seats there; the Conservatives, with only twice as many votes in the West, won twenty-five times the number of seats. In Quebec, the results were reversed: the Conservatives, with 13 percent of the vote, won just one seat; the Liberals, with two-thirds of the Quebec vote, won nearly all the seats there. The Gazette (Montreal), Mar. 1, 1980. The result was to make the Liberals appear to be—and to behave as—a wholly Eastern party and the Conservatives a wholly Anglophone party, thereby exacerbating ethnic and regional tensions. The remedy proposed is to supplement first-past-the-post seats with a minority of seats elected by provincial constituencies on a PR formula. The aim is to assure parties seats wherever they have significant votes and so to mitigate polarization. Alan C. Cairns, "The Strong Case for Modest Electoral Reform in Canada" (unpublished paper presented at the Harvard Univ. Seminar on Canada-United States Relations, Sept. 25, 1979). The regional distribution of party support was quite different, however, in the 1984 parliamentary elections.
Structural Techniques to Reduce Conflict 65 1
intractable situation. Various combinations of electoral formula, ballot structure, and constituency delimitation can operate to preserve fluidity, promote moderation, induce coalition, and produce proportionality.
None of these effects is a matter of party reconstruction: no system examined here abolishes ethnic parties any more than it succeeds in fragmenting them. Many of the effects are matters of party posture more than structure. A party may remain ethnically based but become more moderate. But nothing in ethnic conflict in civilian regimes is more crucial than whether a party adopts a conciliatory posture on ethnic issues.
Similarly, the work of electoral systems is vulnerable to the offsetting effects of other variables: the strength of the conflict, the way territory has been carved up, the timing of innovation (did the proliferation of states, for example, precede the new electoral formula?). The electoral system is one part of the total framework of incentives and disincentives in which ethnic groups and parties operate. The Nigerian experiment shows with supreme clarity the extent to which countervailing incentives can produce mixed outcomes. What stands out, in spite of the limitations, is just how important a piece of the incentive structure the electoral system is and what a dearth of imagination there has been in most countries in utilizing its potential for ethnic accommodation.
THE SUBSTITUTABILITY OF TECHNIQUES
The analysis of territorial and electoral innovations makes clear that the political incentive structure is one package. We have seen that the operation of some pieces of that structure can offset results that might have been achieved by other parts of the structure. By the same token, there may be more than one way to achieve the same result. In the Nigerian federal system, it will be recalled, the proliferation of states helped bring to the fore subethnic divisions that catalyzed party formation in Northern Nigeria, thereby enhancing inducements to multiethnic party support. Lebanon reached a similar destination by a quite different route. By insisting on mixed tickets and reserved offices, the Lebanese electoral system required cooperation across ethnic lines, thereby channeling conflict in intraethnic directions. There is, then, a degree of substitutability of techniques.
The implications for policymakers are clear. A range of means is avail-
652
Ethnic Groups in Conflict
rStanr^TrU£U'ar CndS- if the Pr°SPect of solution arouses resistance, as ,t does ,„ many countries, electoral change may seem I threatening. If there is fear that a regime may tinker with the ele«o system ,n its own narrow interests, territorial realignment may prove more prom.s.ng course. Substitutabil.ty enhances the Pol>cymakers m the quest for conflict reduction.
CHAPTER SIXTEEN
Preferential Policies to Reduce Ethnic Conflict
Electoral policies have a short time frame from adoption to impact; preferential policies do not. A new electoral formula or constituency delimitation produces (or fails to produce) a political response in the next election or two. Any reduction in conflict behavior on the part of politicians will soon be visible. Policies to reduce ethnic disparities, by augmenting the representation of particular groups in, for example, modern-sector employment or higher education, do not have the same quality. Measuring changes in ethnic representation is more difficult than computing election results, and representational changes do not necessarily augur changes in ethnic conflict behavior. Then, too, the impact of preferential policies on ethnic relations may vary over time, compounding the difficulty of evaluation.
All of this may simply be to say that electoral and, for that matter, territorial changes do not really operate on the structure of society but on the epiphenomenal behavior of voters and politicians. Preferential policies have more far-reaching aims and are inherently more difficult to guide to their intended destination. The difference resembles the distinction developed by Martin Shapiro and James Q. Wilson between point and line decisions. "A point decision is one that involves a self-effectuating choice among competing alternatives; a line decision is one that requires the coordination by plan of the actions of many people extending over a substantial period of time." With point decisions, the consequences are "immediately and continuously felt."' With line decisions,
1. James Q. Wilson, "What Can Be Done?" (unpublished paper presented at the fourth annual Public Policy Week conference sponsored by the American Enterprise Institute, Washington, D.C., Dec. 10, 1980), 23.
653