Party Politics and Different Paths to Democratic Transitions

Benin and Senegal represent two successful cases of democratic transition in Africa. They also represent two different paths to that end. This article explores the role of political parties in facilitating these different and successful paths to democratic transitions. In Benin, political parties and political leaders relied on the prevailing patterns of ethno-regional cleavages to structure their strategic interactions, mobilize electoral support and organize competition in legislative and presidential elections. In Senegal, an incremental pattern of institutional reform helped the ruling party retain power while enabling fragmented opposition groups to participate in competitive elections. In the long run, this helped opposition groups develop an effective electoral coalition to defeat the ruling party in presidential and legislative elections and bring about a turnover in government.


Introduction
Benin and Senegal represent two successful cases of democratic transition in Africa. They also represent two different paths to that end. In Benin, the transition originated in popular demonstrations against the incumbent authoritarian regime, followed by a broadly representative national conference in which new democratic institutions were negotiated, and culminated in founding legislative and presidential elections. In Senegal, the transition followed an evolutionary path in which incremental institutional reforms enabled an erstwhile authoritarian incumbent and the ruling party to retain power even while allowing opposition parties opportunities to participate in competitive elections, leading eventually to the defeat of the ruling party by an opposition coalition. This article examines the role of party politics in facilitating these different but successful paths to democratic transitions.
In Benin, examination of the results of presidential and legislative elections since the advent of democracy in 1991 reveals that voting behavior is heavily influenced by ethno-regional cleavages (Bako Arifari, 1995;Banegas, 1998). The Beninois political parties and the party system remain, and in the foreseeable future are likely to remain, structured along ethnoregional lines because ethno-regional groups and cleavages serve as costeffective resources for group mobilization and interest definition, and the resulting strategic interactions of ethno-regionally-based leaders animate the competition for political access and representation. This in turn suggests that politics in Benin rests on consociational expectations about the distribution of public goods, expectations that political parties help to structure by their organization and articulation of ethno-regional interests and to validate by their success in securing public goods for their ethno-regional supporters.
Scholars have typically characterized Senegal as a 'pseudo' (Diamond, 1999: x) or a 'quasi' (Vengroff and Creevey, 1997;Villalon, 1994) democracy, but the success of the perennial opposition party leader, Abdoulaye Wade, in winning the presidency in 2000 and of his party, Parti Démocratique Sénégalaise (PDS), in winning a majority of seats in the National Assembly in 2001 indicates that Senegal has made a further transition towards full democracy. In fact, although the presidency is the crown jewel of Senegal's political system, the electoral reforms that facilitated its historic turnover were first introduced for national legislative and regional and local elections. The interconnections of these multiple-level elections explain the choice of electoral rules and the ensuing pattern of democratization and transfer of power in Senegal (Mozaffar and Vengroff, 2002).
Two very different types of parties and party systems grounded in distinct social structural configurations shaped these different processes of democratic transition. In Benin, an authoritarian regime that was imposed in 1972 after 12 years of political instability was institutionalized through the rule of a single party, the People's Revolutionary Party of Benin (PRB), and survived until 1990 when widespread popular protest precipitated by deepening economic and financial crises led to the convening of the National Conference. From this conference emerged the institutional framework for democratic elections and a liberal democratic state (Decalo, 1997;Heilbrun, 1993). Political parties re-emerged with the onset of democratic transition, but, replicating the pattern witnessed during the post-World War II nationalist struggles for independence from colonial rule when political parties first emerged, most quickly came to reflect the country's ethnoregional cleavages that serve as cost-effective sources of strategic information, interest articulation and collective action. In Senegal, an evolutionary process of electoral reform, party change and shifting coalitions led eventually to the defeat of the dominant ruling party, the Parti Socialiste (PS), and turnover in government. In both countries, however, party leaders and voters engage in a complex coalition-building game that shifts over time by the strategic entry and exit of different political parties. In both countries, moreover, the intricate strategies adopted by party leaders to maintain electoral support and acquire political power have been carried out in a context where voters and politicians appear to know and accept the rules of the game. This acceptance underpins the success of democracy.

Ethno-regional Cleavages, Political Parties and Democratic Transition in Benin
In the widely representative National Conference that mediated the democratic transition in Benin, grassroots organizations, trade unions, religious forces, opposition groups and representatives from the incumbent authoritarian regime negotiated the institutional design of the new democracy. The principal features of this design were: (a) a unicameral legislature with 64 seats allocated by the proportional representation (PR) Largest Remainder-Hare formula in six multi-member constituencies comprising the country's six administrative districts, 1 and (b) an executive president directly elected by a two-round majority formula whereby, if no candidate receives an absolute majority of votes in the first round, the top two finishers compete in a second round of voting. In the ensuing founding legislative and presidential elections held in 1991 under these institutional provisions, the PRB was roundly defeated in legislative elections and a former World Bank administrator Nicéphore Soglo defeated the authoritarian incumbent Mathieu Kérékou in the presidential elections. Since then, Benin has held three more legislative and three more presidential elections that are generally assessed to be free and fair. 2 Two key features characterize these elections. The first is the rapid proliferation of political parties that is characteristic of most post-authoritarian democracies. In Benin, 79 parties officially registered after the political liberalization in 1991 (Degboé, 1995: 75-6). By 1998, that number had increased to 118. While not all registered parties compete in the elections, a substantial number do. For example, 30 individual parties or alliances competed in the 1991 transitional elections, while 31 individual parties and alliance parties competed in the 1995 elections. An important reason for this proliferation of parties is that, because of the political restrictions under authoritarian rule, political actors participating in competitive elections in post-authoritarian democracies suffer from severe information deficit about the extent of their electoral support and the appropriate campaign strategies for mobilizing it. Lacking such information, candidates and parties typically rely on familiar sources of information to structure their strategic interactions and coordinate over votes and seats. In this context, ethnicity and associated patterns of ethnic cleavages and spatial dispersion of ethnic groups serve as important sources of such information (Mozaffar et al., 2003;Mozaffar and Scarritt, this issue).
Thus the second key feature of elections in Benin is the decisive influence of ethno-regional cleavages on voting patterns, party strategies and the structure of party systems. 3 This influence, however, is not reflexive. It is contingent on variations in the morphology and spatial dispersion of ethnic groups in Benin, combined with the design of electoral systems for legislative and presidential elections adopted in 1991. This contingent influence of ethnic cleavages and the interaction of these cleavages with the electoral system prevent any single ethnic group from imposing its political domination, motivating instead the formation of political coalitions as the principal means of winning competitive elections. Three factors, specific to the configuration of ethno-regional cleavages and electoral systems design, encourage and facilitate this strategy of coalition-formation in Benin.
First, the size and internal cohesion of ethnic groups do not correspond perfectly to their concentration in the broadly defined northern and southern regions of the country. Major ethnic groups with large population shares, such as the Fon in the south and Bargu in the north, are internally divided along various sub-ethnic dimensions, creating both inter-ethnic and intra-ethnic cleavages, while those that are internally cohesive comprise a small percentage of the population. Second, ethnic groups and subgroups are also spatially dispersed in one or more of the six administrative districts of the country -Atacora and Borgu in the north and Atlantique, Mono, Ouémé and Zou in the south -although a group or subgroup may be spatially concentrated within each administrative district. 4 Third, as noted above, these administrative districts comprise the electoral constituencies in which political parties compete for national legislative seats allocated by the PR formula and for votes in presidential elections decided by the two-round majority formula. Thus, the complex group morphology and spatial distribution of ethnic groups, both of which preclude any single ethnic group from winning electoral majorities on its own, combine with the institutional designs of the electoral system to exert strong pressure for political parties to form electoral coalitions that cut across the characteristic inter-ethnic and intra-ethnic cleavages and the corresponding group and subgroup distribution across electoral constituencies.
A useful way to measure the influence of ethno-regional cleavages on voting patterns is to compare the percentage of nationwide votes with the percentage of regional votes won by political parties. Table 1 reports these percentages for the 1991 legislative elections in which two major parties and four major alliances won a combined 77 percent of the national vote reported in the last row of the table. The two parties were: (1) Rassemblement National pour la Démocratie (RND), which won 12.08 percent of the A comparison of the vote distribution disaggregated by constituencies reveals that each party and alliance receive substantially more votes in their respective ethno-regional strongholds in each constituency than their national vote shares. Except for two cases, the ethno-regional vote percentages of the parties and alliances exceeded their national vote percentages. The MNDD/MSUP/UDRN alliance in Atacora registered the highest vote difference of 30 percent, followed by a 26 percent difference for the PSD/UNSP alliance in Mono, and a 20 percent difference each for the UTR/UFDP/ULD/MDPS alliance in Atlantique and the NCC in Ouémé. The exceptions are the PRD/PNDD alliance, whose national vote share (17.27 percent) exceeded its ethno-regional vote share (12.44 percent), and the RND, whose national (12.08 percent) and regional (12.53 percent) vote shares were almost the same.
That no party or alliance registered a majority of votes in its ethnoregional stronghold attests to the contingent effects on voting patterns and party support of the complexity of ethnic group morphology associated with politically salient inter-group and intra-group cleavages as well as the spatial distribution of the groups and subgroups within and across individual constituencies. In other words, an ethnic group or its subgroup may be dispersed across two or more constituencies, as for example the Fon across Atlantique and Zou constituencies. But the Goun, a subgroup of the Fon, are concentrated largely in Ouémé, while the Aizo, another Fon subgroup, are concentrated in another part of Atlantique. 5 And to the extent that the spatial concentration of different ethnic groups moderates the direct effect of differences among them on voting patterns and party support (e.g. Mozaffar et al., 2003), the combination of inter-group and intra-group differences interacting with associated patterns of spatial concentration results in the dispersion of votes among several parties within the same constituency. Moreover, use of the PR formula at the constituency level creates incentives for maintaining group cohesion and supporting parties that represent group interests at that level, which facilitates inter-group and intra-group bargaining and the formation of electoral coalitions, but also leads to the dispersion of votes between the various parties and coalitions in the constituency and thus to the observed plurality instead of majority winning margins. Table 2 reports the results of the 1995 legislative elections, for which 83 seats (increased from 64) were allocated in 18 multi-member constituencies created by dividing the existing 6 administrative districts which had served as electoral constituencies for the 1991 elections into 3 multi-member constituencies each, with district magnitude ranging from 3 to 9 seats. The national results show that, consistent with the pattern of strategic entry and exit of many smaller parties and the resulting pattern of temporary electoral alliances in each election noted by Mozaffar and Scarritt in their contribution, a different configuration of eight political parties captured a total of 61 percent of the national votes: (1) the PRD running on its own won 14.24 percent of the national vote; (2) the new Renaissance du Benin (RB), which was created by the first lady Rosine Sogolo to support her husband who could not rely on the fragile political coalition that helped him win the presidency in the transitional election of 1991, won 13.45 percent; (3) the PSD without an alliance partner won 7.31 percent; (4) a new electoral alliance, the Front d'Action pour le Renouveau et le Développement (FARD-ALAFIA), won 6.96 percent; (5) the NCC, running again on its own, won 5.65 percent; (6) the new Union pour la Démocratie et la Solidarité Nationale (UDS) won 5.07 percent; (7) the new Rassemblement des Démocrates Libéraux (RDL) won 4.63 percent; and (8) the old Parti Communiste du Bénin (PCB) won 3.92 percent.
A comparison of the ethno-regional and national vote percentages of these parties reveals the pattern of ethno-regional influence witnessed in the 1991 elections. The new FARD-ALAFIA alliance registered the widest difference of 70 percent between its ethno-regional vote percentage of 76.77 percent and its national vote percentage of 6.96 percent, winning pluralities in all three constituencies of its ethno-regional stronghold of Borgou. The PSD also registered a wide difference of 54 percent, receiving 61.20 percent of its vote in Mono compared to its share of 7.31 percent of the national vote. The NCC exceeded its ethno-regional percentage over its national percentage by 44 percent in Ouémé (49.73 percent to 5.65 percent). PRD did the same by 40 percent in Ouémé. The newcomer RDL received 45 percent and 16 percent if its vote in Borgou and Atacora, respectively, compared to its national percentage of 4.63. Finally, the UDS, another newcomer, received 5.07 percent of the national vote, but 38 percent of its vote in Borgou and 14 percent in Atacora.
This analysis of the 1991 and 1995 election results suggests that the combination of (a) ethnic group morphology featuring both inter-group and intra-ethnic cleavages, (b) the associated pattern of spatial distribution of ethnic groups and subgroups within and across electoral constituencies, and (c) the use of the PR formula to allocate seats in multi-member constituencies creates incentives for party leaders to maintain group solidarity, mobilize electoral support and negotiate electoral coalitions at the constituency level in legislative elections. Presidential elections produce a similar strategic dynamic. Because the district magnitude of presidential elections is one, and because of use of the two-round majority system to determine the winner, weaker candidates representing smaller configurations of ethno-regional support have an incentive to enter the first round and display their control of blocks of votes that they can use to enter into strategic bargains with the top two first-round winners who compete in the second round. Stronger candidates also have an incentive to enter into such bargains because the combination of inter-group and intra-group cleavages in their ethno-regional strongholds does not guarantee the outright majority support needed to win presidential elections, increasing the cost of mobilizing sustained support across elections. The second round results of the presidential elections reported in Table  3 show that the winners in both elections did well in their ethno-regional strongholds. In the 1991 elections, Kérékou won 32.42 percent of the national vote, but won a whopping 93.61 percent of his national vote in Atacora and 96.43 percent in Borgou. In the 1996 elections, Kérékou won 54.00 percent of the national vote, but again won an overwhelming 93 percent of the vote in Atacora and 85.50 percent in Borgou. Sogolo registered a similar wide difference in his national and ethno-regional vote percentages. In 1991, while winning 67.57 percent of the national vote, he won 93.19 percent in Atlantique and 90.67 percent in Zou. In 1996, he lost the presidential election with a national vote percentage of 46 percent, but won 70.30 percent of the national vote in Atlantique and 80.10 percent in Zou.
The data in Table 3 also reveal that presidential candidates in Benin have to rely not only on forging and retaining support in their ethno-regional strongholds, but also on building cross-regional alliances in order to win. A close examination of the data shows that the dramatic 35.15 percent increase in Kérékou's share of the national vote that led to his victory in 1996 was due almost entirely to the shifting cross-regional alliances he was able to build. In both Mono and Ouémé, for example, Sogolo's former allies in the 1991 elections decided to join Kérékou, which is reflected in the increases in his vote share of 37.81 percent in Mono and 59.75 percent in Ouémé. Kérékou was also able to forge cross-regional alliances with several regional candidates, such as Adrien Houngbédji (PRD) and Sacca Kina (FARD-ALAFIA), and improve his vote margins in Sogolo's ethnoregional strongholds. In Atlantique his vote share increased from 6.81 percent in 1991 to 29.70 in 1996, and in Zou from 9.33 percent to 19.90 percent.

Political Parties and Democratic Transition via Incremental Institutional Reform in Senegal
Senegal, like Benin, is a multi-ethnic society, but ethnicity has not played the same role in politics in Senegal as it has in Benin. Instead, particularly in the immediate pre-and post-independence years, the opinion leaders were heads of the Muslim brotherhoods, and especially of the Mouride brotherhood. Although the Mouride brotherhood was primarily Wolof, the largest ethnic group constituting about 43 percent of the population, it did not emphasize or use ethnic loyalty as the basis or claim to its right to direct its followers. The French, who had established the tradition of supporting Mouride leaders and not interfering with brotherhood affairs in return for Mouride leaders' support of colonial rule, also did not play ethnic groups off against each other. The government of Leopold Senghor, which took over Senegal at independence, followed the same pattern. In succeeding years, political opposition and inter-party rivalry were based on struggles for ascendency between followers of different elite politicians, all of whom claimed to be working for ideological goals such as modernization, the end of corruption, and the rights of the poor. The major exceptions to this nonethnic pattern of Senegalese politics were the periodic protests by ethnic groups in the Casamance, who did claim that the needs of their region and their ethnic groups were being ignored. These struggles, although sometimes violent, have not been central to politics in Senegal, however.
In the case of Senegal, an evolutionary process of electoral reform, party change and coalitional configurations led to an eventual turnover of power over a period of three decades. In February-March 2000, Senegal held a historic two-round presidential election. On 19 February, Abdou Diouf, the incumbent and leader of the long-ruling hegemonic party, the PS, won a plurality but not the required majority in an eight-candidate race. In the run-off held on 19 March, Senegalese voters gave a resounding victory to Abdoulaye Wade, the veteran opposition leader, ending the 40-year dominance by the PS. This event was followed by the adoption of a new constitution in a national referendum held on 7 January 2001 that established a new republic with an institutional structure modeled on France's semipresidential system. The National Assembly, still under majority control of the PS, was dissolved on 15 February 2001, and a period of rule by presidential decree, as specified under the new constitution, lasted for nearly three months. The National Assembly elections of 29 April 2001 completed the march toward a complete and total transfer of power from the PS to the liberal democratic PDS and the so-called 'SOPI' (change) coalition.
The PS was the vehicle for control by the first president of Senegal (Leopold Senghor) and his chosen assistants. This party rose to prominence because Senghor was astute enough to see that political power could not be based on the small urban population. He built a coalition of party support that crossed ethnic and regional lines and had as its base the support of the Muslim leaders and, in particular, the head of the powerful Mouride brotherhood (Behrman, 1970). Thus Senghor's Bloc Démocratique Sénégalais (BDS) was able to absorb several of Senegal's smaller parties and become the dominant party by 1958, when it was renamed the Union Progressiste Sénégalaise (UPS). The stage was now set for the emergence of the de facto, but never de jure, one-party state at independence from France in 1960 (Creevey and Morgenthau, 1984;Nzouankeu, 1984).
A series of ensuing electoral reforms animated a process of incremental democratization. Analytically, these sequential reforms are best understood in terms of their strategic multi-level relationships. For the UPS (renamed the PS in 1976), electoral reforms for the National Assembly elections improved opportunities for retaining control of the presidency. Reforms of rules governing local elections to rural and municipal councils were explicitly and implicitly tied to the reforms at the presidential and national legislative levels. The fragmented opposition parties accepted these rule changes because they provided strategic opportunities not only for access and representation, but also for eroding the power of the hegemonic party and wresting control of the presidency (Mozaffar and Vengroff, 2002).
Mounting social pressures engendered by economic failures and a high level of unrest in the rural areas forced the ruling PS to undertake political reforms. In 1974, political regulations were relaxed allowing a new opposition party, Parti Démocratique Sénégalais (PDS), to be officially registered. Again in 1976, under continuing demand for political liberalization, Senghor had the constitution amended to make the country a 'de jure' threeparty system. Legal recognition was given to three parties, each designated as representing one of the 'main' ideological tendencies. The PS was designated as social democratic, the PDS as liberal democratic and the Parti Africain d'Indépendence (PAI) as Marxist-Leninist. This meant that the officially prescribed ideological space was filled. 6 This limited opening effectively eliminated the growing threat to Senghor and the PS by his erstwhile ally, Cheikh Anta Diop, and his party, the Rassemblement National Démocratique (RND), which was excluded since it did not fall anywhere in the officially designated ideological space. It is at this stage that democratic electoral reform begins to play a significant role, eventually generating a life of its own.
The local and municipal elections were also held in 1978 under the winner-take-all system, with one-third of the seats on the rural councils reserved for the cooperatives, long dominated by the Muslim brotherhoods. The support of this powerful group of local elites was thereby secured for the PS, thus enabling 'democratic' elections with limited opposition to pre-empt incipient factionalism from weakening the all-encompassing hegemonic party. The results of the election produced a strong showing for Senghor, who received 82 percent of the vote compared to 17.4 percent for his first-ever opponent, Abdoulaye Wade of the PDS. In the National Assembly, the PS won 82 seats and the opposition PDS 18 (the PAI was shut out). The seemingly low 63 percent turnout was claimed by the RND as a sign of its broad support, since it had urged voters to stage a boycott. The democratic opening in the electoral system cost the PS some seats in the National Assembly and control over a single municipal council, 7 but it legitimized a relatively unfair presidential election in which the media and other government resources were employed to ensure a favorable outcome for President Senghor. The opposition, especially the PDS, gained respectability and took on the role of the first official opposition. As an official party group (15 seats were required to attain this status), the resources available to the PDS increased dramatically, and it had an entrée and a base from which to promote further reform and its own cause.
Senghor officially stepped down in 1981, turning over power to his constitutionally designated successor, Prime Minister Abdou Diouf. Recognizing the potential threat from an opposition united behind Wade, Diouf had the constitution amended (against Senghor's wishes) to eliminate the limits on the number of legal parties. The 1983 electoral reforms were designed to consolidate Diouf's power over the PS, create opportunities for the opposition to win some National Assembly seats and legitimize the presidential succession at the polls. The reforms introduced a mixed (parallel) system with an increase in National Assembly seats from 100 to 120. Half the seats were allocated by PR on a national list to attract the opposition and half by plurality bloc vote at the departmental level to ensure a majority for the PS in the Assembly. This latter formula, one of the least proportional electoral systems (Katz, 1997) assured the PS of a clear majority in the Assembly. The 20 additional seats, all allocated by plurality at the department level, enabled local party 'patrons' to select their favorite candidates and increase the likelihood that the PS deputies who had lost seats in the previous election would have a good chance of returning to power. Finally, permitting party competition but with legal proscription of inter-party electoral alliances assured a high degree of fractionalization of the opposition while providing a sense of democratic legitimacy for the presidential contest. The ballot structure, which required voters to cast single votes for a party list that would then be aggregated to calculate both the plurality and PR seat allocations, reduced the possibility of coordination among the opposition parties. Coordination on plurality seats would reduce opportunities to win the PR seats which the weakly organized opposition parties were better positioned to gain.
The results of the 1983 elections were, as expected, highly favorable for the PS. The opposition won only 9 of the 120 seats, even though its proportion of the vote increased slightly over that of 1978. The PDS was reduced to eight seats, a number too small even to be recognized as an official parliamentary group. The RND, charging voting fraud, refused to accept its one seat and was replaced by a dissident factional representative of the party. In the presidential race, Diouf, having successfully consolidated his power over the PS, faced five opposition candidates and easily won 85 percent of the vote, a total higher than that obtained by Senghor in 1978. The turnout declined to 58 percent, but with Anta Diop participating this could no longer be attributed to the strength of the opposition boycott. The same electoral system was retained for the 1988 elections, but only six parties took part, as opposed to the eight in 1983. A declining economy and increased unemployment, especially among the youth, created areas of strength for the PDS in the urban centers, particularly Dakar (Young and Kanté, 1992: 67). Even though the electoral system and the advantages of incumbency (e.g. control of state media) continued in favour of the organizationally strong PS, the results cut into the PS majority. Diouf's official vote share declined to 73.2 percent and the PDS increased its share of parliamentary seats from 8 to 17 while winning 43 percent of the vote in the capital (Kanté, 1994). The PS retained its overwhelming legislative majority with 103 seats. Most significantly, the strong showing by the PDS, violent protests against alleged election fraud, opposition boycotts of the 1990 local elections and growing factional battles in the PS forced Diouf to invite Wade and three of his PDS colleagues to join a short-lived government of national unity. This action restored regime legitimacy and provided the opposition with resources, leverage and status that were hard to pass up. The price for Diouf was additional reform of electoral rules for the National Assembly election.
The short-lived coalition government, the majorité présidentielle élargie, produced an important set of political reforms with the involvement of all political parties. These included a new electoral code, representation of all parties at polling stations, a guaranteed secret ballot, a lowered voting age (from 21 to 18), an easier and expanded system of voter registration, guaranteed access to the state media for all parties and the acceptance of foreign election monitors (Kanté, 1994). Additionally, one of the most important concessions to the opposition changed the balance in the allocation of seats, increasing the number of PR seats decided by a national list from 60 to 70 and decreasing the plurality seats by 10 to 50. In the tworound majority presidential elections, a candidate was now required to secure the support of at least 25 percent of the registered voters in order to win in the first round. The opposition felt that this would increase the possibility of going to a second round where they might unite behind a single candidate. Despite harboring lingering doubts about its implementation, the opposition parties strongly supported the new system as a sound basis for free and fair elections.
The PS leadership fully understood that the party would incur some losses in the National Assembly, but remained confident of a majority because of its advantage in the plurality races. The increased number of proportional seats encouraged opposition parties to participate but also reduced their incentives to unite, while the 25 percent requirement for winning the presidency attracted many presidential candidates in the first round. In the face of mounting internal and external pressures for democratization, bringing the opposition back in after its boycott of the 1990 local elections would bolster the legitimacy of the presidential results. The presidential term was extended from 5 to 7 years, thus separating presidential and parliamentary elections in the future. Rising factionalism among the ruling elite was also undermining the PS. This factionalism was engendered by the attempt of key leaders to improve their access to Diouf and hence their strategic positions in the anticipated succession struggle. Expanding the importance of the national list thus strengthened the hand of the president while weakening the regionally based factions that exercised greater influence over the local level plurality seats.
The 1993 presidential and National Assembly elections were held under the new rules. Table 4 displays these results and the results of subsequent elections to show the shifting patterns of vote distribution between the PS and PDS, and the associated shifts in electoral coalitions that eventually led to the PDS victory over the PS in the 2000 presidential and 2001 National Assembly elections. In the 1993 National Assembly elections, PS with 57 percent of the votes won 84 seats (70 percent), while the PDS with 30 percent of the votes won 27 seats (23 percent). Among the several parties, the Alliance Jëf-Jál (AJ) with 5 percent of the vote and the Ligue Democratique (LD) with 4 percent of the vote won 3 seats each, the Parti de l'Independence et du Travail (PIT) with 3 percent of the vote won 2 seats, and the Union Démocratique Sénégalaise Renovation (UDSR) with 1 percent of the vote secured 1 seat. In the presidential elections, President Diouf was re-elected with 58 percent or slightly more than 30 percent of the electorate, a comfortable but slimmer than expected margin over the 25 percent required. The PDS candidate, Wade, won 32 percent of the vote, with 6 other candidates sharing the remaining 10 percent. The battleground for electoral reform now shifted to the local level in preparation for the 1995 local elections (Vengroff and Ndiaye, 1998). The 1990 municipal and rural community (C-R) elections were conducted on the basis of party lists with a single area-wide multi-member district and a plurality electoral formula, virtually ensuring total PS control of all seats on any given municipal or C-R council and prompting the weak and fragmented opposition parties to boycott the elections. In 1995, however, the opposition parties targeted the local councils to strengthen their political support in preparation for the National Assembly elections in 1998 and the presidential elections in 2000. After a rancorous debate, and a 1-year delay, significant reforms in the electoral system were implemented and the opposition opted to participate. The revised electoral code introduced a mixed electoral system that virtually assured some representation for opposition parties in local governments.
In the 1996 local elections held under the revised code system, half of the seats were allocated on the basis of a bloc vote with a plurality formula in a city-or C-R-wide multi-member constituency, as had been the case for previous elections but for all seats. The other half of the seats were allocated on the basis of a PR formula (as in the National Assembly elections in 1983 and, in a modified form, in 1988) using the Hare quota. The code required party-list voting in a single round and proscribed vote transfers. The size, and hence the district magnitude, of both the municipal and C-R councils was significantly increased. In addition, in the Dakar Region, a major opposition stronghold, the existing three councils were broken up into 43 communes d'arrondissement, making it more difficult for the opposition to compete effectively and assuring that the opposition would win only part of the capital.
In fact, the PS retained dominant control of the local councils (losing its majority in only 7 of 98 municipal and 18 of 316 rural councils) although the opposition won nearly 2,500 seats (out of about 13,000) in the various councils across the country. The relative fairness of these elections also gave the PS a better sense of its actual electoral prospects. In particular, it highlighted the threat that the ongoing intra-party factionalism and declining electoral support nationally posed to its continued domination of the National Assembly, especially if the distribution of 70 PR and 50 plurality seats for the 1993 elections were retained for the 1998 elections. However, opposition acceptance of the formula for the 1996 local elections legitimized the return to a 70 plurality and 70 national list PR national election as well.
The 1998 electoral reforms also specified that each of the 30 departments must have at least 1 plurality seat and that no department would exceed 5 seats. Thus, although the allocation was supposed to be consistent with population, the district magnitude limits favored the smaller regional centers. Dakar and Pikine, already with 5 seats each and both opposition strongholds, were not awarded any of the 20 additional plurality seats, despite their heavy population concentrations. The seats were instead allocated to departments that heavily favored the PS, contributing to the party winning 18 of the 20 new seats. The overall results favored the PS, whose seat shares changed marginally from 70 to 67 percent even though its vote shares dropped from 57 percent in 1993 to just slightly over 50 percent in 1998. Its total number of seats actually increased from 84 to 93. Thus, the results of the 1996 local and the 1998 National Assembly elections were virtually identical, validating the justification of new electoral rules for the 1998 national elections in terms of the new electoral rules negotiated by the PS and the opposition for the 1996 local elections.
In 1997, after intense lobbying by the opposition, an all-party conference established the independent Observatoire National des Elections (ONEL). Significantly, at the conference, President Diouf played the role of a mediator between the PS, which opposed the creation of ONEL, and the opposition parties, which saw ONEL as a restraint on the PS. Working closely with the politically mobilized non-government organizations throughout the country, ONEL played an effective role in legitimizing the results of the 1998 elections.
The changes in electoral rules in 1998 again strengthened President Diouf's hand for the 2000 elections while attracting, but also fragmenting, the opposition. Diouf's ability to nominate candidates for the additional 20 seats reinforced his authority while adding to the pool of party faithful. The ballot structure required a single vote for a party that applied to both the plurality and the proportional seats. Hence, opposition parties had an incentive to present separate slates in as many constituencies as possible. The accumulation of all votes cast at the department level for national list allocation reduced their prospects of coordination by selectively competing for plurality seats in some areas. 8 The 1998 electoral reforms also enabled Diouf to create an upper house, the Senate, in the national legislature. The highly attractive salaries and other perks provided for the new senators made these positions extremely attractive in and of themselves and as sources of patronage for the party's 'barons'. A special electoral college composed of local, municipal and regional officials and National Assembly deputies selected most of the senators. Twelve seats were reserved for presidential appointment. This combination of elected and appointed seats produced an exclusive PS Senate that owed loyalty to the president. The opposition, however, raised serious questions about its existence and expenses. The successor government headed by Wade has now abolished it in a constitutional referendum.
The close vote margin and low voter turnout in the last two presidential elections posed the real threat that the PS presidential candidate might have to compete in a second round against a united opposition, especially if the first-round victory required the support of 25 percent of the registered voters. In the summer of 1998, the constitution was amended to eliminate the requirement linking a first-round majority victory to any fixed proportion of the registered voters. Thus, even a very low turnout would have no affect on whether a second-round presidential election would be required. Finally, party deposits that were forfeited by candidates who failed to attract 5 percent of the vote were doubled to 6 million CFA, a move that would keep some of the smallest parties out of the race.
These reforms helped to weaken the solidarity of the dominant coalition. The PS, always a loose grouping of personalist factions, was shaken by the defection of several key leaders. Djibo Ka, the former Minister of the Interior, left the PS and established the Union pour le Renouveau Democratique (URD), which promptly won 11 seats and finished third in the 1998 National Assembly elections. Subsequently, Moustapha Niasse, a foreign minister under Diouf, also left the PS and formed his own party, the Alliance des Forces du Progrés (AFP) for a run at the presidency. He has support from some sections of the Islamic brotherhoods. Bearing in mind that the PS was only able to muster half of the vote in the 1998 legislative elections, this important defection posed a major threat to Diouf's ability to win a first-round victory in the 2000 presidential election. With Djibo Ka and Niasse's respective factions deserting the PS and the center-left parties uniting behind Abdoulaye Wade of the PDS, the 2000 presidential race became extremely competitive. The opposition denied Diouf, a first-round majority victory and for the most part unified behind Wade to defeat him in the second round.
The election of Abdoulaye Wade as president succeeded because of the support he was able to garner from a coalition with the AFP and the support of several smaller parties of the so-called 'pole de gauche' (Vengroff and Magala, 2001). In the inter-party accords, the coalition agreed to support the candidacy of the opposition leader who qualified for the second round of the presidential elections against the PS candidate, Abdou Diouf. The inter-round election accord was for Niasse to be named prime minister in the event of an opposition victory in the presidential elections. After the election, Wade honored the agreement and appointed Niasse as prime minister. But it should be noted that Senegal has always had a strong presidential system, in spite of the series of constitutional changes over time that created, abolished, and recreated the office of prime minister. The prime minister, appointed by the president, had very tenuous links to the National Assembly and served basically as a tool of the president. Any hint that the prime minister was developing an independent power base was quickly dealt with either by replacement or by suppression of the office. During the election campaign, Wade had talked about moving to a system of power-sharing between the PM and president. Precisely how to implement the move away from pure 'presidentialism' had to be elaborated constitutionally.
The immediate difficulty facing the coalition was that the legislative elections in 1998 had produced a strong PS majority in the National Assembly.
Plans to reform the system, eliminate corruption, and move in new directions institutionally had to be put on hold because the necessary institutional reforms could not easily be pushed through the National Assembly without major concessions to the PS. Wade thus sought to reform the system through the mechanism of a new constitution. Serious debate took place as to whether the presidential system in place should be supplanted by a pure Westminster-style parliamentary system or a semi-presidential system modeled on the French Fifth Republic. Although Wade had talked about the advantages of a parliamentary system, once he was in the presidential palace the issue of how to produce a stable ruling coalition capable of initiating and implementing serious reform took precedence.
Given the pragmatic necessity of holding the new governing coalition together, the semi-presidential option was chosen. The logic of the choice seemed to reflect the need to keep the Niasse forces (the AFP) on board for a constitutional referendum, which would be consistent with a continued power-sharing arrangement between President Wade and Prime Minister Niasse. The new constitution, with the endorsement of most political parties, was approved overwhelmingly in a referendum in January 2000 that witnessed a 66 percent turnout and 94 percent voting in favor of reform. The promulgation of the new constitution gave the president the authority and opportunity to dissolve the National Assembly and call for new elections (Article 105). In addition, once the assembly was dissolved the constitution gave the president the right to rule by decree in the interim period with the advice of the Council of State (Article 106). He thus had the authority to decide the nature of the electoral system to apply to the upcoming (2001) legislative elections. All parties were consulted so as to legitimize the process.
The electoral system in Senegal has been modified numerous times in past decades. Most of the changes have been designed to make the system more democratic and open to some opposition gains while maintaining the ruling party's margin of seats as a control factor (Mozaffar and Vengroff, 2002). Wade, as leader of the opposition, had always argued for more proportionality in the system and less reliance on the plurality bloc vote, which heavily favored the party in power. For a number of years, Senegal had relied on a mixed electoral system. The smaller parties and the opposition have always argued for greater numbers of seats to be allocated on the national list, while the ruling party has always favored a balance -but one which ensured its domination of the plurality seats. For the 2001 elections, Wade, who had bitterly attacked the machinations by the PS, was in a position to alter this inequitable formula, a solution much preferred by the smaller parties in his coalition.
Once in control of the presidency and with the right to rule by decree, including the power to modify the electoral system, Wade could install a system designed to be more representative of the voters' wishes or to maximize the opportunities for his supporters. Significant numbers of leaders deserted the PS and opportunistically moved into the PDS fold. Coupled with the greater resources now available to the PDS, including the full weight of the presidency, this made the highly inequitable system the PDS had previously attacked suddenly seem attractive. The PS and the AFP, the only other large parties, also felt that they could profit from an emphasis on the plurality side of the seat distribution by finishing first in several department blocs. They therefore chose to argue for increased weight for the plurality side of the election. The smaller parties pushed for pure proportionality based on a national list or some compromise that would provide greater opportunities for a better seat-to-vote distribution. The PDS opted to decrease the National Assembly seats from 140 to 120 and move from a 70-70 split of seats between plurality and PR allocation sides to a 65-55 split in favor of plurality. The PDS calculated that as the new party in power it could win a plurality in many departments, thus increasing its share of seats relative to its vote share.
These calculations proved correct in the 2001 legislative elections. Although the PDS received just under half of the votes (49.6 percent), it won 89 of the 120 seats (74.2 percent). The former ruling party, the PS, finished second in terms of votes with 17.4 percent but garnered only 10 seats, all on the PR list. The AFP finished third in vote shares with 16.1 percent and passed the PS in seats with 11, two of which it won on the plurality side by finishing first in one department. The URD, with 3.7 percent of the vote, garnered 3 seats, one of which was by plurality in a small department (the home area of its leader). The AJ/PADS, with just over 4 percent of the vote, won only 2 seats, both on the national list. Five additional parties were awarded one seat each on the national list by virtue of having the largest remainders, even though they did not achieve a full quota in votes. Some leaders of the SOPI alliance tried unsuccessfully to deny these parties their seats, but the electoral law and its allocation procedures were clear on this issue. The remaining 15 parties, which presented lists, were shut out of the seat allocation because of very low vote totals.
The disproportionality in this election greatly exceeded even the high rates Senegal had experienced under PS rule. In the 1993 elections, in which 70 seats were allocated by proportional formula on a national list and 50 by department on plurality vote, Gallagher's least-squares disproportionality index in the allocation of seats and votes was 11.2. This figure rose to 12.5 for the 1998 elections, in which seats were allocated 70-70 between the 2 electoral formulas. The index jumped to 19.3 in the 2001 elections. Thus, although coming to power on the basis of arguments for democratic reform, the PDS-SOPI coalition of parties has taken a major step backwards in this regard with its manipulation of the electoral system to its own advantage.
In order to consolidate his power base, Wade dismissed his increasingly independent-minded coalition partner, Niasse, and appointed the country's first female prime minister, who was a government functionary, lacked an independent political base and was thus beholden to the president. He later replaced her with Idrissa Seck, a close political ally who was serving as director of the President's Office and Minister without portfolio. Once Seck became widely seen as Wade's likely successor Wade replaced him with a less threatening former Minister of the Interior, Macky Sall. In effect, his selection again consolidated virtually all power in the hands of the president. During the campaign, Wade tried to make the election a referendum on his leadership and change. He argued forcefully for the need to have a unified leadership and against the threat of a government in which he lacked a strong legislative majority. Having gained 89 of the 120 legislative seats, Wade's PDS and the ever-changing SOPI coalition (now comprising the PDS, LD, AJ and URD) are now left in control, with a very weak and divided opposition led by Niasse's AFP and the remnants of the PS. Overall, the one-party dominant system itself did not change dramatically. The effective number of legislative parties, measured by the Laakso-Taagepera index, was reduced to 1.76 from 2.10 after the 1998 elections. This occurred in spite of the fact that the effective number of electoral parties had increased from 2.4 in 1993 to 3.22 in 1998 to 3.29 in 2001.

Conclusion
Senegal and Benin represent two different paths to democratic transition, but both paths underscore the importance of political parties in the transition. In Benin, the progress toward democracy has not been continuous. An authoritarian regime lasted from 1972 to 1990. Since then, democratic elections have been successfully held and leaders of different parties have replaced each other in power. As in Senegal, changes in the electoral rules in Benin have been important, although with a different net effect. In Benin, electoral districts present a high level of ethnic convergence. This explains why many parties generally get most of their votes from very limited areas, but because of the distinct morphology of ethnic groups and their spatial dispersion, cross-regional alliances become essential strategic tools for securing electoral victory.
Party politics in Benin emanates from the strategic interactions between ethno-regionally based political forces. Benin is divided along persistent ethno-regional cleavages that defined political interactions before the imposition of authoritarian 'Marxist-Leninist' rule and re-emerged with the reintroduction of democratic institutions. The structure of these cleavages, reflected in patterns of ethnic group fragmentation and spatial dispersion and concentration, motivates the combinations necessary for survival in the national political arena, as both party leaders and voters find ethnicity to be cost-effective political resources for structuring their strategic coordination over votes and seats and the exchange of public goods for electoral support.
Exemplifying a pattern witnessed recently in Mexico, Taiwan and South Korea, among others, Senegal evolved toward greater democracy, but in a context in which the hegemonic party gradually allowed for the organized expression of opposition to its dominant position. Understanding this political dynamic requires a systemic or a 'whole system' approach that emphasizes analysis of the intricate but often hidden relationships of elections and the rules governing them at multiple levels -presidential, legislative, regional and local. Four factors directly linked to the politics of democratization motivate this approach. First, competitive elections at multiple levels enable authoritarian incumbents to retain power while accommodating opposition demands for access, participation and representation. Second, the choice of rules governing these elections reflects negotiated outcomes to the conflicting goals and interests of authoritarian incumbents and opposition groups. Third, the rule changes produce unintended consequences that tend to aggravate the intrinsic factionalism in hegemonic ruling parties, such as the Parti Socialiste (PS) that governed Senegal for three decades prior to the democratic transition. Finally, the ruling elites' strategic accommodation of opposition demands in the short run ensures their own survival, but in the long run creates opportunities for further democratization and the eventual turnover of power, which is an important measure of successful democratic transitions.
In Senegal, incremental institutional change and the resulting succession of national, regional, local and municipal electoral reforms that were introduced over the years since independence represented efforts to manipulate the electoral institutions to the advantage of the ruling party. Initially, they enabled the PS to consolidate its political domination while allowing the opposition a chance to participate meaningfully. In the long run, however, this strategy undermined the dominant party's control. By the 1990s, as a result, each new reform deepened the existing fissures in the ruling PS, strengthening the opposition forces and steadily moving them toward a minimal winning coalition. These shifts served to promote further democratization and push Senegal to the point where a transfer of presidential and legislative power became both possible and a reality in 2000 and 2001, respectively (Mozaffar and Vengroff, 2002;Vengroff and Magala, 2001).
In both Benin and Senegal, party leaders and voters are engaged in a complex coalition-building game that is altered over time by the strategic entry and exit of parties and candidates. In both countries, the intricate strategies adopted by party leaders to maintain their votes and acquire and hold political power have been carried out in a context where voters and politicians appear to know and accept the rules of the game. It is this acceptance that underpins the success of democracy in both countries, despite some anti-democratic tendencies that still exist in both.

Notes
1 For the 1995 legislative elections, the number of seats was increased from 64 to 83 allocated in 18 electoral constituencies which were created by the division of the 6 administrative districts into 3 electoral constituencies each. Before the 1999 legislative elections, 6 more administrative districts were created and the resulting 12 administrative districts were then divided into 24 multi-member constituencies for allocating the 83 seats in the 1999 and 2003 elections. 2 Since the democratic transition in 1991, Benin has received an average Freedom House score of 2.5 and, therefore, a rating as 'partially free'. 3 This influence is reminiscent of similar influence recorded in the competitive elections before and after independence (Staniland, 1973a, b). 4 For systematic data on the number and size of ethnic groups, their constituent morphology featuring inter-ethnic and intra-ethnic cleavages, and the spatial concentration of each group and subgroup, see Scarritt and Mozaffar (1999). 5 For data on these patterns, see Table 2 in Scarritt and Mozaffar (1999). 6 To further legitimize this system, after the elections, a fourth party, the conservative MRS, was allowed to register in 1979. 7 Oussoye in the Casamance was won by the PDS. 8 Quinones and Vengroff (2002) show that strategic voting in Senegal has declined over time as recognition of the negative impact of doing so on one's true preferences became obvious.