The Institutionalization of Political Power in Africa
Across sub-Saharan Africa, formal institutional rules are coming to matter much more than they used to, and have displaced violence as the primary source of constraints on executive behavior. From decolonization in the early 1960s through the 1980s, most African rulers left office through a coup, assassination, or some other form of violent overthrow. Since 1990, however, the majority have left through institutionalized means—chiefly through voluntary resignation at the end of a constitutionally defined term or by losing an election. While institutional rules may not yet always determine outcomes in Africa today, such rules are consistently and dependably affecting the strategies through which those outcomes are reached.
On 16 May 2006, after months of intense and divisive national debate, the Senate of Nigeria rejected a bill which would have changed that country's constitution to permit President Olusegun Obasanjo a third term in office. By asserting the supremacy of the constitution (with its two-term limit) over the desires of President Obasanjo's supporters that the popular leader be permitted to run for a third term, the Senate's vote marked a watershed in Nigeria's political history. As important as the outcome was the way in which the conflict was resolved—by the votes of duly elected legislators rather than through force or the threat of same. Given that Nigeria's First and Second Republics (1963–66 and 1979–83) were overthrown by military coups, the settlement of this political struggle via the Senate chamber rather than the gun barrel represents a major shift in the way that decisions over executive tenure in Nigeria have been made.1
Both the outcome of Obasanjo's third-term campaign and the process through which it was reached signal a growing trend in sub-Saharan Africa: The formal rules of the game are beginning to matter in ways that they previously have not. Scholarly and popular writers alike have traditionally depicted Africa as a place where formal institutional rules are largely irrelevant. Although every African country has a constitution as well as a body of laws and administrative procedures that place formal limits on executive power, the long-held consensus among observers has been that these rules play little role in actually constraining leaders' behavior. This view is reflected in the "personal rule" or "Big Man" paradigm that has dominated the study of African [End Page 126] politics for the past thirty years. This paradigm's foundational idea is that personal relationships are more important than formal rules and that a leader's decisions will always take precedence over the laws that those decisions might contradict. The field's conventional wisdom has been that rules do not shape leaders' behavior; leaders' behavior trumps rules.
The significance of the Nigerian Senate's actions must be seen against the backdrop of this entrenched view. Contrary to depictions of African politics as "not beholden to formal procedures but to personal decisions," President Obasanjo was forced by the rules of the game (the constitution) to accept something other than his preferred outcome.2 Also, contrary to portrayals of political conduct in Africa as "governed by the awareness that constitutional rules or administrative regulations can, and probably ought, to be evaded,"3 both supporters and opponents of the president's bid for a third term sought to achieve their goals by working through, not around, formal institutional channels. Although these efforts did involve attempts to bribe and otherwise influence the governors and senators who would vote on the third-term issue, what is more important is that the actors whose support was being sought were politicians—people whose influence stems from their position within the formal political system—rather than soldiers or other elements operating outside of it.
Across sub-Saharan Africa, formal institutional rules are coming to matter much more than they used to, and have displaced violence as the primary source of constraints on executive behavior. From decolonization in the early 1960s through the 1980s, most African rulers left office through a coup, assassination, or some other form of violent overthrow. Since 1990, however, the majority have left through institutionalized means—chiefly through voluntary resignation at the end of a constitutionally defined term or by losing an election. Elections are also becoming more important as a mechanism for selecting leaders in Africa, as reflected in the large increase in both their number and their competitiveness. The fact that incumbents still almost always win, however, underscores that the major challenge connected with the task of limiting presidential power in Africa today is not so much promoting elections as making certain that leaders adhere to constitutional limits on their continued eligibility to contest them.
Some African leaders have managed to circumvent restrictions on seeking more than two terms in office, yet have done so through formal institutional channels rather than extraconstitutional means. Thus, while institutional rules may not yet always determine outcomes in Africa today, such rules are consistently and dependably affecting the strategies through which those outcomes are reached. This represents a major change in how power is exercised, and it challenges us to ask again [End Page 127] whether the Big Man still bestrides the world of African politics with as much ease as he once did.
How Leaders Exit Power
During its first decade after independence in 1960, the small West African country of Benin had no fewer than twelve heads of state, every one of whom was overthrown in a coup d'état. This striking record of serial leadership change by force stands in complete contrast to Benin's record since 1990. From that date to 2006, Mathieu Kérékou and Nicéphore Soglo alternated as president following wins and losses in national elections. Moreover, both men have refrained from running for terms beyond those allowed by constitutional limits.4 While Benin provides perhaps the most dramatic example of the change that has taken place in how leaders leave power in Africa, it is nonetheless indicative of a broader trend.
To document this trend, we have collected data on how every African head of state exited power between independence and the end of 2005. Our sample includes some 227 leaders from 46 sub-Saharan African countries.5 We coded each leader's means of exit from office into one of five categories: coup or violent overthrow (including civil war); assassination; natural death; voluntary resignation; and losing an election. We further grouped these categories into two broader classes of cases: those in which leaders left power through regular means (which include natural death, voluntary resignation, or losing an election) and those in which leaders were removed by irregular means (coup or violent overthrow, or assassination). Figure 1 presents the decade-by-decade averages.
As Figure 1 makes clear, nearly three-quarters of the African leaders who left power in the 1960s and 1970s did so through a coup, violent [End Page 128] overthrow, or assassination (depicted by the dashed line). In the 1980s, this dropped to just below 70 percent, and by the 1990s it was surpassed by the share of those who left power through natural death, voluntary resignation, or electoral defeat (depicted by the solid line). Between 2000 and 2005, the share of leaders leaving power through irregular means dropped to just 19 percent. Whereas the modal means by which heads of state in Africa left office used to be coup or assassination, it is now voluntary resignation, in most cases triggered by constitutional term limits (9 of the 17 cases between 2000 and 2005).
Another way of looking at this transformation is to compare Africa with the rest of the world. To do this, we use data from the Archigos project, which codes the way that heads of state in every independent country in the world entered and exited power between 1875 and 2004.6 This comparison, seen in Figure 2, reveals a remarkable degree of convergence. With respect to how leaders leave power, Africa (depicted by the solid line) used to be truly a place unto itself—a place befitting its own theories about politics and power.7 Yet by the first years of the current century, Africa had joined the rest of the world. Whereas African leaders were two to three times more likely than leaders elsewhere in the world to leave power by violent means in the 1960s, 1970s, and 1980s, they are equally likely—or to be more precise, equally unlikely—to leave power under violent circumstances today.8
These trends point to the increasing institutionalization of political power in Africa. Whereas political power used to change hands principally through violence—at a time and in a manner chosen by coup plotters—it now changes hands principally in accord with institutional rules. Of course, the manner in which executives leave office is only one [End Page 129] indicator of how beholden they are to formal constraints in their exercise of power more generally. It says nothing, for example, about the extent to which they adhere to objective procedures when they allocate jobs, award contracts, enforce regulations, or exercise other powers of office. Nonetheless, whether a leader departs office via regular as opposed to irregular means is critical. A regular departure means that there is an understood set of basic limits on how long a head of state may stay in power, as well as on how his opponents may seek to replace him. It therefore marks the most important step toward restraining executive power and institutionalizing political authority more broadly.
One of the clearest manifestations of the increasing institutionalization of executive power in Africa is the increasing importance of elections. Elections have been held in Africa since the independence era, albeit sometimes only intermittently and with varying degrees of contestation. Even so, both the total number of elections held per decade and the share of elections that are meaningfully contested have risen over time, particularly since the early 1990s.
In the 1960s and 1970s, sub-Sarahan Africa saw an average of 28 elections per decade. That number climbed slightly to 36 in the 1980s, and then shot up to 65 in the 1990s, a decade that began auspiciously with the national-conference movement in Benin and the beginning of the end of apartheid in South Africa. From 2000 through the end of 2005, 41 elections had already been held in Africa, and thus this upward trend appears to be continuing. This pattern is largely a product of two developments. First, countries such as Malawi that did not hold elections in the immediate postindependence period have begun to hold them. Second, countries such as Togo that once held sporadic elections have begun to regularize their electoral processes.9
Elections are also becoming more intensely contested.10 In only two of the 26 presidential elections held in Africa in the 1960s did the incumbent actually face an opponent. The vast majority of presidential elections during this period were little more than plebiscites or grassroots-mobilization exercises in which the head of state stood no risk of losing power. By the 1990s, however, more than 90 percent of presidential elections were contested, and by the 2000–2005 period, this share had risen to 98 percent. This dramatic change reflects the growing recognition by African leaders that, to maintain their legitimacy in the eyes of both their own citizens and the international community, they must subject themselves to elections in which opponents have at least a theoretical possibility of winning.
Permitting a challenger to run, however, is not the same thing as putting oneself at real risk of losing power. In many cases, African leaders who bowed to popular or international pressure to hold contested elections found ways of rigging them so that the contests never brought a meaningful risk that the incumbent would be unseated. Nonetheless, [End Page 130] our data suggest that such rigging is becoming harder to accomplish. When we compare across decades the reelection rates of presidents who permitted challengers to run against them, we find that elections in Africa are not only becoming more often contested but also more robustly competitive. During the entire period between 1960 and 1990, only one African president lost an election—Aden Abdullah Osman of Somalia, who was defeated by challenger Abdirashid Ali Shermarke in 1967.11 Since 1990, the loss rate of incumbents has risen to a modest but nonetheless meaningful 14 percent (incumbents lost 14 times in 100 opportunities). African presidents today are thus more than twice as likely to lose power if they subject themselves to contested elections than they were before 1990, when the loss rate was just over 6 percent (just a single electoral defeat in 16 contested elections).
Despite this trend of increasing competitiveness, however, the more important point to underscore is that African leaders who want to stay in power are usually able to do so, even if they allow competitive elections. Incumbent presidents in Africa today still win reelection more than 85 percent of the time. The advantages of incumbency in Africa are so great that elections alone—even free and fair elections—are not enough to put meaningful limits on presidential power. The institutionalization of political power thus depends less on whether sitting presidents are willing to permit challengers to run against them than on whether incumbents will stand down (thereby forgoing likely reelection) when they have completed as many terms as their respective countries' constitutional laws allow.12
Since 1990, more than three dozen African countries have adopted new constitutions, the vast majority of which have included prohibitions against a president serving more than two terms. Figure 3 sorts the region's countries into five categories depending on four factors: 1) whether or not their post-1990 constitution puts a two-term limit on the presidency; 2) whether at any point between 1990 and 2005 that term limit had been reached; 3) whether, if reached, an attempt was made to amend the constitution to overturn the limit; and 4) whether that attempt succeeded.13
As Figure 3 shows, eighteen African presidents during this period have found themselves in the position of having completed two terms and being constitutionally barred from seeking a third.14 These incumbents have had three options: 1) abide by the constitutional term limit and stand down; 2) attempt to change the constitution to permit a third term; or 3) scrap the constitution altogether and prolong their tenure through extraconstitutional means. No African leader has thus far taken this third course, which indicates just how much has changed in the [End Page 131] region. In the 1960s and 1970s, it was commonplace for leaders who wanted to perpetuate their rule to have themselves declared "president for life"—as did the Central African Republic's Jean-Bédel Bokassa (who later promoted himself to "emperor"), Equatorial Guinea's Francisco Macías Nguema, Ghana's Kwame Nkrumah, Malawi's Hastings Kamuzu Banda, Togo's Gnassingbé Eyadéma, Uganda's Idi Amin, and Zaire's Mobutu Sese Seko. Today that option would appear to be closed.
All eighteen directly elected presidents who faced term limits heard strong calls from their supporters to find a way to stay in power. Nine of these chief executives—Kérékou of Benin, António Monteiro of Cape Verde, Jerry Rawlings of Ghana, Daniel arap Moi of Kenya, Alpha Konaré of Mali, Joaquim Chissano of Mozambique, Miguel Trovoada of S~ao Tomé and Príncipe, France-Albert René of Seychelles, and Benjamin [End Page 132] Mkapa of Tanzania—resisted these appeals and announced that they would abide by their countries' constitutions and refrain from seeking a third term. It is unclear whether this was because these leaders feared that they lacked the votes needed to change the constitution, were wary of the concerted opposition that they would face, worried that they might lose a third election, or simply believed that abiding by the constitution was the right thing to do. The answer almost certainly lies in a combination of these considerations and varies from case to case.
Whatever the rationale may have been, the point to emphasize is that stepping down was not necessarily these presidents' most preferred outcome. They agreed to relinquish power because the constitutional prohibition on extending their rule raised the cost of staying in power beyond a level that they were willing to bear. Indeed, several leaders (Moi, Chissano, and Mkapa) made initial moves in the direction of seeking a third term only to change course when it became clear that powerful coalitions were ready to oppose them. Such voluntary relinquishment of power in the face of formal rules telling them that their time was up directly challenges the caricature of Africa as a place where "abstract constitutions and formal institutions exist on paper, but they do not shape the conduct of individual actors, especially those in power."15
Still, the nine leaders who agreed to step down represent just half the universe of sitting presidents who faced term limits. The other nine tried to change their respective constitutions to make a third term possible. Presidents Frederick Chiluba of Zambia, Bakili Muluzi of Malawi, and Obasanjo of Nigeria were rebuffed in their efforts. Chiluba's attempt to secure a third term was undermined by a groundswell of public opposition from civil society groups and a deep split within his own party. Fifteen senior members of the National Executive Committee of the ruling Movement for Multiparty Democracy (MMD) publicly opposed his bid to amend the constitution, and 59 MMD lawmakers, including several cabinet ministers and his own vice-president, signed a document in which they pledged to block any effort to allow him to run for a third term. In the face of such solid opposition, Chiluba backed down. Muluzi also faced strong public opposition to his bid for a third term. Even so, he very nearly succeeded: An amendment that would have abolished term limits fell just three votes short of the necessary two-thirds majority in the legislature.16 As discussed earlier, Obasanjo's bid for a third term failed in the Nigerian Senate.
In six other cases, leaders facing term limits were able to change the constitutional provisions that prevented them from running for reelection and continuing their rule. Presidents Idriss Déby of Chad, Omar Bongo of Gabon, Lansana Conté of Guinea, Samuel Nujoma of Namibia, Gnassingbé Eyadéma of Togo, and Yoweri Museveni of Uganda all succeeded in changing their constitutions to allow themselves the ability to compete for third terms, and all six won their ensuing elections [End Page 133] handily. In Gabon, Namibia, Togo, and Uganda, the constitution was amended by an act of parliament. In Guinea, the amendment came through a national referendum (which third-term supporters won with 98 percent of the vote). In Chad, the change was achieved through both mechanisms: A two-thirds parliamentary vote in favor of amending the constitution triggered a national referendum on the question, which was carried by a two-to-one margin.
These cases remind us that many African leaders still possess the power to shape outcomes to suit their preferences, even when those preferences conflict with formal limitations on what they are legally permitted to do. Presidents Déby, Bongo, Nujoma, and Eyadéma were able to change their countries' constitutions because their respective parties controlled more than two-thirds of the seats in parliament. President Museveni lacked the supermajority required to change the constitution, but was able to use his control of state resources to buy the parliamentary votes he needed to pass the third-term amendment—just as Nigeria's Obasanjo tried to do, unsuccessfully. In contrast to the cases described earlier, these examples would appear to vindicate the view of Africa as a place where leaders monopolize political and economic power so completely that their preferences do in fact take precedence over the formal rules of the game.
Yet this conclusion ignores these rulers' decision to use their considerable powers to work within, rather than around, institutionalized channels. Each of these leaders was probably strong enough to have voided his country's constitution and declared himself president for life, as many previous African leaders once did. Yet each of these more recent presidents felt the need to spend considerable political and financial capital trying to secure the votes to change the constitution—a feeling that in itself shows how heavily the formal rules have come to weigh. Even in the six cases where the rules and their boundaries were circumvented, the very presence of the rules shaped the strategies that the leaders pursued to achieve their favored outcomes.
It is true that earlier leaders also sought to legitimate their seizure of power through institutional means: The declaration of a one-party state in Ghana (and effectively, a life presidency for Nkrumah) was achieved through a 1964 national referendum, and Banda was named Malawi's life president in 1970 by a legislative act. Yet these earlier leaders were only superficially working through the system. It would have been unthinkable for them to fail to win their referendum or necessary legislative supermajority. This differs from the cases since 1990, in which leaders have not enjoyed such undisputed power. Indeed, it was precisely the anticipation of rebuke that prevented leaders such as Chiluba and Muluzi from pushing the third-term issue further, and that led Obasanjo to respond to the Senate's vote by declaring that he would respect its verdict. The roles that referenda and legislatures played in [End Page 134] sanctifying life presidencies in the 1960s and 1970s were little more than formalizations of the facts on the ground rather than, as has been the case since the 1990s, a means of establishing new facts.
Explaining Third-Term Outcomes
The institutionalization of executive power in sub-Saharan Africa depends heavily on third-term issues. Thus an important question to ask is why certain term-limited executives decide either to step down or to attempt to amend the constitution in order to extend their tenure. One potential factor is public opinion. Popular leaders who believe that they will win reelection presumably have a greater incentive to try to change the constitution than those who have reason to doubt their electoral prospects. The contrasting examples of Kenya's Moi and Namibia's Nujoma illustrate this logic. Moi was elected to his second term in 1997 with only 40 percent of the vote and decided not to try to change the rules to allow himself a third term in 2002, while Nujoma, who began his second term in 1994 with 76 percent of the popular vote, decided to push for an amendment that would allow him to run for a third term in 1999. Among the eighteen African presidents who faced term limits between 1990 and 2005, the median margin of victory in the previous election among the nine presidents who decided to try to seek third terms was 41.5 percent. Among the nine who did not, it was only 17.8 percent.
A second potentially relevant factor is a leader's expected benefits of staying in power. It seems a fair assumption that younger leaders—who can expect to spend more years benefiting from office and also more years when they might be out of power and worried about prosecution—will have a stronger urge than older rulers to hang on to power. Our data appear to bear this out: At the time that their second terms expired, the nine presidents who faced term limits between 1990 and 2005 and who attempted to secure a third term had a median age of 60, whereas the median age of the nine who chose to stand down was 66.
A third factor focuses not on the value of staying in power but on the cost of trying to change the constitution. Given the strong emphasis that Western donors have put on the rule of law since the end of the Cold War, the ability of donors to impose their will in a given country should affect whether or not that country's president will try to alter the rules to permit a third term. To the extent that a country's reliance on foreign aid indicates its susceptibility to donor pressure, we should expect presidents in countries that receive high levels of aid relative to GDP to be less likely to seek third terms than presidents in countries that receive low levels of aid relative to GDP. Indeed, the median level of aid in the nine countries in which presidents attempted to secure third terms was 7.3 percent of GDP, whereas it was 12.1 percent of GDP in the nine countries where presidents did not. In keeping with this [End Page 135] pattern, the presidents of the only two significant oil exporters in the sample, Nigeria and Gabon, were among the group that attempted to change the constitution to remove third-term limits.
With just eighteen cases to explore, this analysis can only be suggestive—indeed, none of the differences we have highlighted are statistically significant in a two-tailed t-test. Nonetheless, in addition to providing potential explanations for what has happened in the past, these factors also provide a set of expectations about what may happen in countries with formal term limits that have yet to be reached (see the list in Figure 3). If a leader's popularity matters, for example, then we would expect Rwanda's Paul Kagame—who was elected to his second term in 2003 by a margin of 91 percent—to be more likely to push for a third-term amendment than Liberia's Ellen Johnson-Sirleaf, who will not face a third-term limit until 2018 but who finished second in the first round of the election that brought her to power in 2006.
Similarly, if a leader's age is important, then we would expect Cameroon's Paul Biya and Senegal's Abdoulaye Wade to be more likely to step down when their second terms expire (they will be 81 and 88 years old, respectively) than Burkina Faso's Blaise Compaoré or Madagascar's Marc Ravalomanana, who will be in their early sixties when their second terms end. Finally, if donor leverage matters, then we would expect Johnson-Sirleaf and Sierra Leone's Ahmad Tejan Kabbah, whose countries are heavily dependent on foreign aid, to be much less likely to try to surmount constitutional term limits than Paul Biya or, when his second term ends in 2016, Congo-Brazzaville's Denis Sassou-Nguesso, since both their countries are major oil producers that receive little foreign assistance. Tracing what happens in these countries when the third-term issue eventually presents itself will put us in a better position to assess the causal weight of each of these factors.
The suggestion that donor pressure may be at least partly responsible for the decisions of African leaders who face third-term limits raises the critical question of whether the changes that we have documented with respect to the more general institutionalization of political power may be epiphenomenal—a product of outside forces rather than internal change. Yet even if donor pressure (or the broader international norm that leaders who come to power by means other than elections will not be viewed as legitimate) has contributed to making the rules matter, it remains true that the rules do matter much more than they have in the past. This realization directly contradicts some of the central assumptions so often heard about African politics.
A Case for Cautious Optimism
In 1982, Robert Jackson and Carl Rosberg wrote that "the inquiring student of African politics may be better advised to read Machiavelli or [End Page 136] Hobbes than the 'constitutions,' official plans, or party programs of most African governments if he wishes to understand their central characteristics and dynamics." They saw "little sign that the drama of personal rule will soon give way to more settled institutional forms of conducting the affairs of states."17 Nearly twenty-five years later, Goran Hyden maintained the same position when he wrote that, in Africa, "the notion that constitutional norms and principles are binding on political leaders is still very much in doubt."18 Yet recent evidence suggests that the personal-rule paradigm that has been used to understand African politics for the past thirty years needs to be rethought.
By highlighting the growing importance of formal rules and their constraining effect, we would like to suggest a shift in emphasis in African political studies from its current preoccupation with classifying regimes to a focus on the factors that constrain executive power. An astute reader may have noticed that the word "democracy" has not appeared in this essay. The omission has been deliberate. In the past twenty years, there has been much debate about whether various African regimes should properly be viewed as "pseudodemocracies," "façade democracies," "hybrid democracies," or in terms of some other label that connotes democratic imperfection. Such efforts, we believe, risk diverting attention from the more basic issue of whether or not the behavior of political actors is constrained and, if so, whether by rules or by violence or the threat of violence. A focus on the degree to which formal rules matter (or the degree to which formal rules have displaced informal ones—a topic treated by Michael Bratton elsewhere in this issue19 ), will lead to research and typologies that better capture the salient characteristics of African politics today.
We are making broad claims, and should offer a few cautions and clarifications. First, we are not saying that African leaders today have any less desire to stay in power than their predecessors did; our claim is simply that leaders today are more constrained by formal rules in trying to achieve their most preferred outcomes. They accept electoral defeats when they might prefer to stay in office. Many of them (roughly half) stand down in the face of two-term limits when they would prefer to run for a third term. Or they change the rules so that their preferred outcome no longer violates those rules (in which case the impact of the rules is felt not in the outcome that is reached but in the way it is achieved).
As a final caution, we remind the reader that there are important exceptions to these general trends. There have been recent coups in the Central African Republic, Mauritania, and S~ao Tomé and Príncipe, as well as reversals in the institutionalization of political power in the Gambia, Zimbabwe, and Uganda (as Andrew M. Mwenda shows elsewhere in this issue20 ). Leaders who have ruled for decades remain in power in Equatorial Guinea and Gabon. Furthermore, one could make a compelling case that reversals are not out of the question in several of [End Page 137] the countries that fit the trends we have described—including Nigeria, the country with which we began our discussion. It is therefore important to temper whatever optimism might arise from the arguments and evidence we have presented with the recognition that, despite these general trends, not every African country is moving in a direction of greater institutionalization of political power, and that reversals remain possible in some of the countries that currently appear to be heading toward greater institutionalization.
Nonetheless, in the past fifteen years there has been a substantial change in African politics. It is entirely possible that the power of formal constraints over some African rulers may decline, or that the frequency of coups d'état may again rise, but the lines in Figure 1 are extremely unlikely to recross. In order to grasp this new reality we need to adjust our theories. African politics needs to be viewed through a lens that recognizes the formal constraints on executives and rejects the assumption that African leaders simply get what they want.
Daniel N. Posner is associate professor of political science at the University of California–Los Angeles (UCLA) and author of Institutions and Ethnic Politics in Africa (2005).
Daniel J. Young is a doctoral candidate in political science at UCLA and is finishing a dissertation on political parties in Africa based on fieldwork in Malawi.
1. For a detailed discussion, see Richard L. Sklar, Ebere Onwudiwe, and Darren Kew, "Nigeria: Completing Obasanjo’s Legacy," Journal of Democracy 17 (July 2006): 100–15; and Jibrin Ibrahim, "Nigeria’s 2007 Elections: The Fitful Path to Democratic Citizenship," USIP Special Report No. 182, January 2007.
2. Although Obasanjo never publicly declared his desire for a third term, the effort to extend his tenure was widely understood to have his strong endorsement. The quoted phrase is from David K. Leonard and Scott Straus, Africa’s Stalled Development: International Causes and Cures (Boulder, Colo.: Lynne Rienner, 2003), 3.
3. Robert H. Jackson and Carl G. Rosberg, "Personal Rule: Theory and Practice in Africa," Comparative Politics 16 (July 1984): 425.
4. Kérékou first took power following a coup in 1972. After nearly twenty years in power, he agreed to hold an election in 1991, which was won by Soglo. Soglo then permitted elections to take place five years later, as prescribed by the constitution, and stepped down when Kérékou won. In 2001, Kérékou easily won reelection in a runoff after Soglo and Parliament Speaker Adrien Houngbédji decided to boycott the second round, citing irregularities. In 2006, Kérékou stepped down after agreeing not to attempt to change the constitution, which prohibited him from running both because of a two-term limit for the presidency and a maximum-age limit of 70 years (Kérékou was 72). Soglo, who was 71 at the time, also announced that he would abide by the constitutional age constraint and not run, and the presidential election was won by Yayi Boni.
5. For a similar recent data collection effort, see Arthur A. Goldsmith, "Predatory versus Developmental Rule in Africa," Democratization 11 (June 2004): 88–110. We exclude Swaziland because it is a kingdom in which all leadership appointments are made by royal succession. [End Page 138]
6. Hein Goemans et al., Archigos: A Database on Political Leaders, version 1.4, University of Rochester and University of California–San Diego, November 2004.
7. If we restrict the non-African sample to developing nations, the pattern remains the same: Through the 1990s, African leaders were nearly twice as likely to leave power by irregular means as leaders in other non-OECD countries.
8. There is a difference in the share of leaders leaving power through irregular means in the 2000–2005 period in the Archigos data set (5 percent) and our own (19 percent), which we attribute to differences in the coding rules applied. But since the key comparison in Figure 2 is between Africa and the rest of the world (and uses only Archigos data), this difference across data sets is less important than consistency in coding across regions within the Archigos data set, which we assume to be high.
9. A third development has been the increase in the number of independent countries in Africa, which rose from 36 in the 1960s to 44 in the 1970s, 45 in the 1980s, and 47 in the 1990s. Yet even if every newly independent country held elections at the same rate as those countries already in existence, these changes would not account for the increase we observe in the number of elections over time.
10. We define a contested election as one in which an opposition candidate is permitted to run against the incumbent.
11. Seewoosagur Ramgoolam lost power in Mauritius following an election in 1982. Mauritius has a parliamentary system of government, however, and Ramgoolam lost his post of prime minister when his party lost its majority in parliament rather than through a direct election, so we do not count this as another case.
12. For an analysis of the importance of term limits on a global basis, see Gideon Maltz, "The Case for Presidential Term Limits," Journal of Democracy 18 (January 2007): 128–42.
13. In Figure 3, we exclude Africa’s five parliamentary countries—Botswana, Ethiopia, Lesotho, Mauritius, and South Africa—because they do not directly elect their presidents. We also exclude Comoros (because of its rotating presidency), Somalia (because it has lacked a clear executive during the period under study), Swaziland (because it is a kingdom), Eritrea (because it has never held national elections), and Sudan (because of ambiguity about which constitution should apply). We code Zimbabwe as not having a constitution that provides a two-term limit because the clause that provided for such a limit was changed before 1990, the starting point of our analysis. Similarly, President Mugabe’s efforts to extend his current term through 2010 are taking place after the end point of our analysis.
14. President Compaoré of Burkina Faso could be argued to represent a nineteenth case. His opponents called for him to be barred from running in 2005 on the grounds that he would, by that time, already have served two elected terms in office. The Constitutional Council, however, ruled that the two-term limit, which had been suspended in 1997 but was reinstated in 2000, did not apply retroactively and that Compaoré’s term of office should be counted from the time of the most recent amendment. Consequently, we code this as a case where a term limit has not been reached.
15. Goran Hyden, summarizing the personal-rule paradigm, in African Politics in Comparative Perspective (New York: Cambridge University Press, 2006), 98.
16. Muluzi actually pushed for two separate amendments—first, for an "open [End Page 139] term" (abolishing term limits altogether), and then for an amendment allowing the incumbent president to stand for a third term. It was the vote for the former bill that very nearly passed. With declining support for Muluzi’s bid, the latter was with-drawn before being put to a vote. It is also worth noting that Malawi’s constitution specifies the limit only as two consecutive terms, and at the moment Muluzi’s party, the United Democratic Front, looks poised to nominate him to run in the 2009 presidential election.
17. Robert H. Jackson and Carl G. Rosberg, Personal Rule in Black Africa (Berkeley: University of California Press, 1982), 266.
18. Hyden, African Politics in Comparative Perspective, 111.
19. Michael Bratton, "The Democracy Barometers: Formal versus Informal Institutions in Africa," Journal of Democracy 18 (July 2007): 96–110.
20. Andrew M. Mwenda, "Personalizing Power in Uganda," Journal of Democracy 18 (July 2007): 23–37. [End Page 140]