Incumbents in electoral regimes are often able to retain power despite having to regularly compete in multiparty elections (Gandhi and Lust-Okar Reference Gandhi and Lust-Okar2009; Knutsen, Nygård, and Wig Reference Knutsen, Nygård and Wig2017). A growing body of research on democratic erosion suggests that incumbents can entrench themselves in power through constitutional engineering directed at undermining institutions that would otherwise constrain them (Albertus and Menaldo Reference Albertus and Menaldo2018; Levitsky and Ziblatt Reference Levitsky and Ziblatt2018; Prempeh Reference Prempeh2008). Incumbents can also resort to illicit tactics, such as jailing opposition leaders or stuffing ballot boxes, to prevent voters from exercising their democratic rights (Cheeseman and Klass Reference Cheeseman and Klass2018; Simpser Reference Simpser2013).
In this article, we identify a distinct channel through which incumbents can contain the threat of multiparty competition, without institutional manipulation or electoral malfeasance. We argue that incumbents can prevent the emergence of a strong opposition—one that might dislodge them from power—by simply coopting them. By demonstrating a willingness to trade individual ministerial appointments for temporary political allegiance, incumbents can tempt opposition politicians to create splinter parties or to pursue independent candidacies (Bob-Milliar Reference Bob-Milliar2019; Kadima Reference Kadima2014; Lembani Reference Lembani2014). In the process, incumbents do more than merely buy off individual opposition politicians; they weaken the opposition as a whole by inducing their fragmentation.
We formalize the relationship between cooptation and fragmentation through a game theoretic model. The model shows that, when an incumbency advantage exists in electoral regimes, opposition politicians prefer to run for office independently with the goal of receiving a cabinet post rather than attempting to defeat the incumbent as a united opposition. Incumbents can induce opposition fragmentation by developing reputations for bringing their opponents into government. But this cooptation strategy comes at a cost for the incumbent: by offering a cabinet post to a member of the opposition, the incumbent risks alienating members of the ruling party who expect their own share of patronage. Given such a trade-off, the model shows that weak incumbents are more likely to rely on the cooptation strategy to increase their chances of winning the next election. Strong incumbents have less need to depend on opposition cooptation to secure their reelection.
The model’s formulation helps to account for elite political behavior in electoral regimes where informal patronage relationships continue to shape the jockeying for power despite the adoption of formal democratic institutions that allow multiparty competition under universal suffrage. The model underscores that the advantage incumbents derive from coopting opposition politicians is not in amalgamating those voting blocs with their own; it is in the partisan fragmentation that arises among the opposition. Incumbents can use patronage to encourage politicians to maintain their independent party labels rather than coalescing under a single banner. When incumbents prove willing to allocate offices across partisan lines, their opponents find it more difficult to attract others in forming an alternative coalition to the one in power.
We provide empirical evidence for the relationship between patronage cooptation and opposition fragmentation using original data on presidential elections held in 35 sub-Saharan African countries from 1990 through 2016. African presidents historically used patronage appointments to consolidate elite support for their autocratic regimes in the postindependence period (Jackson and Rosberg Reference Jackson and Rosberg1982; Meng Reference Meng2020; Oyugi Reference Oyugi2006), and their successors have continued to use similar cooptation strategies to contend with the exigencies of multiparty competition (Bleck and van de Walle Reference Bleck and van de Walle2018; Ndegwa Reference Ndegwa2001; Osaghae Reference Osaghae1999). We show that leaders who appoint opposition politicians to their ministerial cabinets face a more fragmented field in subsequent presidential elections. We estimate that appointing any opposition politician to the cabinet is associated with at least two additional presidential candidates in the next election, ceteris paribus. We find that this cooptation effect increases with each additional preelection opposition appointment to the cabinet. Consistent with our formal model, we find that weaker reelected incumbents are more likely to make opposition appointments to the cabinet.
This study offers several contributions to the scholarship on democratic erosion. First, while our study extends existing research on elite cooptation in autocratic settings (Buckles Reference Buckles2019; Gandhi Reference Gandhi2008; Gandhi and Buckles Reference Gandhi and Buckles2016; Lust-Okar Reference Lust-Okar2005), we emphasize that this cooptation strategy fits any range of electoral regimes (democratic or authoritarian) precisely because it does not rely on coercion, repression, or explicit rule breaking. The extant scholarship on backsliding often focuses on illicit or violent tactics that incumbents use to stay in power, but our study highlights how democratic erosion can occur when leaders adapt patronage as a legal means for undermining electoral competition. Incumbents quietly subvert democracy by weaponizing their executive authority to make valuable cabinet appointments.
A second contribution of our analysis is to provide a corrective to the conventional depiction of opposition in transitioning and democratizing countries. By endogenizing opposition weakness rather than assuming it stems from exogenous sources, our study suggests that electoral opposition is not necessarily born weak; it can be deliberately made weak. Patronage appointments by incumbents exacerbate the collective action problem that frequently prevents the opposition from competing effectively in countries where the future of democracy remains in question. Given prior research on the effect of presidential elections on party system development (Elgie et al. Reference Elgie, Bucur, Dolez and Laurent2014; Golder Reference Golder2006; Hicken and Stoll Reference Hicken and Stoll2011), our empirical findings provide new insight into how incumbents can impede the emergence of a stable party system—even when the formal institutional requirements are in place.
Opposition Fragmentation through Patronage
Incumbent leaders worldwide have historically distributed political offices to stave off threats to their power (Huntington Reference Huntington1968; Scott Reference Scott1969). They can induce support for their regimes by exploiting the financial weakness of rival politicians who often need ongoing access to resources to sustain their own followings. The politicians who accept patronage appointments are then not merely neutralized by the incumbent; they are dissuaded from acting in concert with others to change the status quo. Once they have accepted patronage appointments, these politicians now have more to lose. Patronage-based cooptation, in short, enables incumbents to stabilize their regimes by impeding the coordination of broader opposition.
Patronage appointments have been particularly effective in allowing incumbents to subdue their opposition wherever elections have become normalized as part of politics. In eighteenth-century Britain, Robert Walpole overcame the instability that followed the Glorious Revolution by using the growth of executive offices to tame a fractious parliament, thereby engineering the extended period of Whig political dominance (O’Gorman Reference O’Gorman1975). Walpole understood that opposition Tories, facing rising electoral costs along with growing intraelite competition, could not resist accepting his patronage appointments. Indeed, under Walpole’s tenure, “[i]t was patronage that cemented the political system, held it together, and made it an almost impregnable citadel, impervious to defeat” (Plumb Reference Plumb1961, 189).
Patronage-based cooptation, beyond permitting leaders to consolidate their regimes, encourages opposition fragmentation. The association between cooptation and fragmentation is especially likely where politicians can gain access to state resources without being in the winning coalition. In Italy, party system fragmentation has been conventionally attributed to the interaction between electoral rules and ideological polarization. But the electoral calculus of politicians was also driven by the established practice of trasformismo, the cooptation of opposition through the formation of ad hoc parliamentary majorities that distributed spoils across the ideological spectrum (Di Palma Reference Di Palma1977). Recognizing that ideological distance did not impede political accommodation, Italian politicians had little incentive to coalesce during the five decades following the Second World War.
The patronage-based cooptation that facilitated regime stability and opposition fragmentation in historical cases continues to occur in countries that underwent political liberalization at the end of the Cold War. Among African countries, the patronage dynamics established during the era of one-party rule survived the transition to multiparty politics (Khisa Reference Khisa and Amundsen2019; Mwenda Reference Mwenda2007). African presidents have used their discretion over government appointments to exploit the financial vulnerability of opposition rivals who are often expected to satisfy the demands of constituents by engaging in clientelistic outreach. To be viable for office, politicians must acquire the means not only to pay for mundane expenses associated with campaigning but also to offer the resources needed to reaffirm their commitment to redistribute among constituents (Kramon Reference Kramon2016).Footnote 1
The problem for African opposition politicians is that they must fend for themselves in financing their campaigns because the fundraising channels typically employed in established democracies are unavailable to them. While ruling parties tap state coffers to subsidize their candidates, few opposition parties have the campaign war chest necessary to provide any substantive funding to their own candidates (Arriola Reference Arriola2012). The financial constraints of opposition politicians thus leave them vulnerable to patronage cooptation by the incumbent. Lacking other financing options, politicians can trade the electoral support they command for a government appointment, especially a cabinet post, that will then enable them to transform public resources into targeted goods or services for their supporters.
Cabinet appointments provide incumbents with an ideal mechanism for credibly distributing patronage to the cash-strapped politicians they seek to coopt. Unlike other forms of compensation, such as one-time monetary payments, cabinet posts approximate the key factors that Oyugi (Reference Oyugi2006, 63) identifies as motivating alliance formation in African countries: “the need to control power and by extension access to the benefits associated with power holding.” Cabinet posts endow their officeholders with benefits that they can direct at their own discretion such as collecting rents on government contracts, handing out public sector jobs to allies, and targeting state resources to constituents (Bob-Milliar Reference Bob-Milliar2019; Hassan Reference Hassan2020; Kramon and Posner Reference Kramon and Posner2016). By accessing such resources, the politicians who accept cabinet appointments can retain their status as the political leaders of their respective constituencies.
Kenya provides such an example. President Daniel arap Moi was reelected in 1997 by winning only 40% of the vote. He feared that the opposition might coordinate in the National Assembly, where his ruling party held a slight majority. Moi therefore sought to coopt Raila Odinga, an opposition party leader, by appointing him as a cabinet minister in 2001. Odinga’s cabinet appointment had an observable effect on his subsequent campaign spending. Examining data on donations Odinga’s party made to Harambees—mass rallies Kenyan politicians historically used to cultivate electoral support (Widner Reference Widner1992)—illustrates how the entry of an opposition leader into government affects their access to resources. In the 16 months prior to joining Moi’s cabinet, Odinga’s party donated a little over $3,000 per month to Harambee projects across the country. That monthly figure jumped by 80%, to over $5,500 per month, in the 15 months after he became a cabinet minister (Osendo and Gachucha Reference Osendo and Gachucha2003).
An incumbent who begins making opposition appointments to the cabinet effectively signals to others in opposition that they too might be coopted. Opposition politicians can then see themselves having a credible option for accessing state power and resources: rather than trying to win an election through a unified opposition alliance, they can individually negotiate themselves into the incumbent’s government. This is an especially attractive option where presidents enjoy an incumbency advantage in elections due to factors such as the ruling party’s superior campaign funds or use of government resources for campaigning. Under such circumstances, if alternation in power is expected to be difficult for the opposition to achieve—even when united—opposition politicians may opt to direct their efforts toward negotiating their individual cooptation rather than building an opposition electoral coalition.
To fragment the opposition, incumbents need to establish a history of extending patronage across the political spectrum without regard to partisan affiliation. When opposition politicians decide their electoral strategy, they will necessarily consider the incumbent’s record of cabinet appointments to determine whether future cooptation is a possibility. Opposition politicians might not expect to negotiate a patronage bargain with an incumbent who has only appointed members of the ruling party to the cabinet. But these politicians could anticipate negotiating such a bargain with an incumbent who previously appointed opposition politicians as cabinet ministers. Since electoral competition creates a structure for repeated play, politicians can anticipate whether an incumbent is likely to renege on such patronage promises (Myerson Reference Myerson2008). The ability of aggrieved opposition politicians to respond in tit-for-tat fashion by coordinating in the next election, and thereby realizing the very outcome the incumbent had sought to prevent, could be a sufficient deterrent to an incumbent reneging.
The logic of cooptation helps to explain why an entrenched incumbent like Cameroon’s long-serving president, Paul Biya, has appointed opposition politicians to the cabinet despite winning every multiparty election. Biya can outspend his rivals with state funds or use coercion whenever necessary, but the fragmentation of his opposition has been most effectively secured through cooptation. After winning the 1992 presidential election with only 40% of the vote, Biya appointed eight opposition members to his cabinet. Since Biya continued making such appointments after subsequent elections, opposition leaders have had little incentive to coalesce. Biya’s strategy proved effective in fragmenting, and thereby weakening, his opposition over time. Whereas five opposition candidates contested in the 1992 presidential elections (with the median candidate winning 4% of the vote), 22 opposition candidates contested in the 2011 presidential elections (with the median candidate winning less than 1% of the vote).
Even in more democratic settings, incumbents who develop reputations for bringing the opposition into government can induce their rivals to campaign for elections individually rather than unifying. Many in the opposition enter presidential races not because they expect to win outright, but to visibly demonstrate their ability to mobilize votes. After an election, opposition candidates can then leverage those votes to negotiate their entry into government. In Benin, opposition politicians have most often contested presidential elections independently rather than joining broad-based coalitions. They do so, in part, because Benin’s successive incumbents have made it a practice to appoint some of their opponents to the cabinet. When President Mathieu Kérékou ran for reelection in 2001, he did so by competing against one of his own ministers, Bruno Amoussou, a veteran opposition party leader who was then reappointed to the cabinet after placing fourth among 17 candidates. Similarly, in 2011, President Yayi Boni ran for reelection by competing against his own justice minister, who ran alongside 12 other opposition candidates. After the election, Boni appointed a different opposition presidential candidate as justice minister.
The Benin example suggests that cooptation strategy is not risk-free for incumbents. A cabinet appointment does more than provide opposition politicians with access to state resources; it also legitimizes them as potential national leaders. Some opposition politicians may only seek to use a cabinet appointment to maximize their personal benefits, but others could invest those resources into competing against the incumbent in future elections.Footnote 2 In Senegal, after President Abdou Diouf brought opposition politicians into the cabinet, one of those opposition cabinet members, Abdoulaye Wade, was able to leverage his ministerial experience to defeat Diouf in the 2000 presidential election (Kelly Reference Kelly2018). Yet, for most incumbents willing to coopt their rivals, this risk is attenuated because opposition parties often splinter as politicians seek to better position themselves for future cabinet reshuffles. In Malawi, successive incumbents managed to stoke intraopposition divisions through selective cabinet appointments. In the run-up to the 2004 presidential election, defectors from the opposition Malawi Congress Party split off to form two new parties, the Republican Party and the New Congress for Democracy. Those new party leaders were subsequently appointed to the cabinet (Lembani Reference Lembani2014).
Despite the advantages to cooptation, incumbents face constraints in its application. One of the most binding is associated with party institutionalization—namely, the extent to which parties operate as autonomous organizations with stable internal structures and regularized decision-making procedures beyond the personalized influence of their leaders (Meng Reference Meng2020; Riedl Reference Riedl2014). Whereas incumbents may value the flexibility that patronage allows in the formation of political alliances, organizationally robust ruling parties seek to veto patronage appointments that cut into the rewards for their partisans.Footnote 3 In Senegal, when Diouf sought to coopt Wade, ruling party elites blocked the initial offering of a cabinet post that was perceived as too powerful (Mendy Reference Mendy2001). More generally, incumbents should be less likely to deploy cooptation at will under more institutionalized ruling parties.Footnote 4
The constraint on cooptation may also stem from the opposition itself. Politicians may be unwilling to accept a cabinet appointment even when it is offered. Although ideological cleavages can be relatively weak in some countries (Bleck and van de Walle Reference Bleck and van de Walle2018), sharp partisan divides run deep in others (Buckles Reference Buckles2019). Opposition parties that have actively stoked polarization to mobilize voters against the incumbent are more likely to have high levels of cohesion that deter defection (LeBas Reference LeBas2013). Where such divisions exist, it may be personally difficult and electorally costly for a politician to accept a post in the incumbent’s government.
Less clear is how cooptation might be constrained by formal institutions like electoral rules. Incumbents have an incentive to engage in opposition cooptation whenever they face the threat of losing power, regardless of the rules under which they compete. Not only do incumbents want to prevent their opponents from coordinating before an election in a first-past-the-post system, but they also need to do the same in a majoritarian runoff system because a second round automatically creates a focal opposition candidacy around which others can rally. Incumbents who compete in runoff systems thus need to consider whether they can divide their opponents before the first round.
We ultimately expect incumbents across regime types, whether democratic or autocratic, to employ cooptation because it is an adaptable, legal strategy that relies on neither repression nor violence. Incumbents can subvert their competition by simply using their executive powers of appointment. In this respect, an incumbent’s electoral regime only needs to fulfill two scope conditions for the mechanism to operate. First, the regime holds multiparty elections in which opposition parties are allowed to compete. Second, there is a recognized incumbency advantage that lowers expectations of opposition victory even with completely free and fair elections. Both scope conditions shape the model we present below.
A Theoretical Model of Opposition Fragmentation
We model the relationship between incumbent patronage and opposition fragmentation as an infinitely repeated sequence of election cycles involving an incumbent, who remains in office until losing an election,Footnote 5 and two opposition candidates. At the beginning of each election cycle, the opposition candidates each decide to fragment or unify based on their common beliefs about how the incumbent will fill a cabinet position if he wins—namely, whether he will appoint someone from his own party or from an opposing party. These beliefs are formed on the basis of how the incumbent filled the position in the past. We assume that opposition candidates would be willing to accept a cabinet appointment if such a position were offered.
The incumbent is more likely to win if the opposition is fragmented than if it is unified because, in the case of fragmentation, the votes the incumbent does not receive are spread across multiple candidates rather than received solely by the leader of a unified opposition. In each cycle, the election winner decides whether to give a cabinet position to someone outside the party, as opposition patronage, or to someone from his own party. Opposition leaders are more likely to believe that they may receive an offer of a cabinet seat if the incumbent has a history of making outside appointments. This expectation makes opposition candidates more likely to fragment, which in turn increases the incumbent’s victory probability. However, offering a cabinet position to an opposition candidate is also costly for the incumbent because it prevents him from being able to offer that position to someone from his own party or existing ruling coalition.
Model Setup
Consider election cycle t, where each election is a three-period stage game of the infinitely repeated game. At the start of period 1, two opposition candidates appear. Both simultaneously choose either “fragment” (i.e., run separately against the incumbent through two distinct opposition parties) or “unify” (i.e., join forces and run against the incumbent in a single opposition party). These choices are made to maximize their respective payoffs from election t forward, based on beliefs about how the election winner will fill a cabinet vacancy, as explained shortly. If both choose “unify,” then unification occurs in election t. Otherwise fragmentation occurs.
For convenience and to simplify the analysis, we label one opposition candidate the opposition leader and the other the opposition member. The intuitive meanings of these labels only apply in the event of unification, meaning that the opposition leader runs as the sole candidate of the opposition coalition. The opposition member, in contrast, forgoes her opportunity to run as a candidate for that election in exchange for a guaranteed cabinet appointment in the event that the opposition leader wins. If the opposition candidates fragment, the leader enjoys no electoral advantage over the member; they run as separate equals against the incumbent.
Let q t-1 denote a binary variable equaling 1 if the cabinet appointment in the previous election t-1 was from outside the party and 0 if it was from inside the party. Let N t-1 be a binary variable equaling 1 if election t-1 was won by a new officeholder and equaling 0 if it was won by the incumbent. Both opposition candidates observe q t-1 and N t-1 at period 1’s start and consider this information when deciding whether to fragment or unify. Let ft L (q t-1,N t-1) and ft M (q t-1,N t-1) denote the period-1 choices of the leader and member, respectively, where both are binary variables equaling 1 if “fragment” is chosen and 0 if “unify” is chosen. Let Ft (q t-1,N t-1) be a binary indicator equaling 1 if the opposition fragments and 0 if it unifies in election t, so that Ft (q t-1,N t-1) ≡ 1 – (1 – ft L (q t-1,N t-1))(1 – ft M (q t-1,N t-1)).
In period 2, stochastic election results occur. The party receiving the highest vote share wins, and vote shares are independent across elections. Under fragmentation, there are three possible election results (incumbent wins, leader wins, or member wins), whereas under unification there are two possible results (incumbent wins or leader wins). Let pt (Ft (q t-1,N t-1)) denote the incumbent’s probability of victory in election t. We assume ½ ≤ pt (0) < pt (1) < 1—that is, there is an incumbency advantage, and the incumbent is more likely to win against a fragmented opposition than against a unified one. If the incumbent loses, the opposition leader and member win the election with equal probability under fragmentation. The current holder of a cabinet post retires at the end of period 2, creating a vacancy.
In period 3, the election winner fills the cabinet post either with someone inside the party or with someone outside the party. Let Dt (qt ) be the winner’s cost of filling the cabinet vacancy in election cycle t, where Dt (0) = 0 and Dt (1) = d > 0. The cost of making an inside appointment is (normalized to) zero, and the cost of making an outside appointment is d. Footnote 6
If the election winner was the incumbent, he is free to choose either qt = 1 or qt = 0. In the case of fragmentation, if an opposition candidate wins, he is also free to choose either qt = 1 or qt = 0. Here we assume that if the winner under fragmentation is an opposition candidate, she will appoint the other opposition candidate if she decides to appoint a candidate outside the party (i.e., qt = 1). We define winners who have the option of appointing someone from inside or outside the party as “unconstrained winners.”
When the opposition is unified, we assume that if the opposition leader wins she must choose qt = 0 and uphold her commitment to appointing the opposition member to the cabinet post. This contract pertains only to the election cycle in which the newly elected incumbent first enters office; after that, the incumbent can choose either qt = 1 or qt = 0 for as long as she keeps winning. Winners who must choose qt = 0 are called “constrained winners.”
Following the choice of qt , payoffs are received, and the opposition candidates who lost the election and received no cabinet appointment retire at period 3’s end. Postelection payoffs vary across three possible outcomes: winning the election, losing the election but getting a cabinet appointment, and losing the election and getting no cabinet appointment. The election winner receives a payoff of R – Dt (qt ), representing the rents and returns to holding office. R > d is assumed, so that the value of winning office exceeds the cost of making an outside appointment. The candidate who loses the election but gets a cabinet appointment receives a payoff of u, where 0 < u < R, representing utility gained from having a cabinet position. The payoff of losing the election and getting no appointment is normalized to zero.
If the incumbent wins and chooses qt = 0, both opposition candidates get payoffs of zero. If the incumbent wins and chooses qt = 1, and if the opposition is unified, then the opposition leader is granted the cabinet position with its payoff of u, and the opposition member receives zero payoff. If the incumbent wins and chooses qt = 1, and the opposition is fragmented, then each opposition candidate is granted the cabinet post with equal probability: the candidate who gets the post receives a payoff of u, whereas the other candidate gets zero payoff. If the opposition is unified and the incumbent loses, then the opposition leader wins, receives a payoff of R, and honors an enforceable contract by appointing the opposition member to the cabinet position with its payoff of u. If the opposition is fragmented and the incumbent loses, then each opposition candidate wins with equal probability, and the losing opposition candidate gets the cabinet post (with its payoff of u) if qt = 1 and zero payoff if qt = 0. The winning opposition candidate is assumed not to give the cabinet post to the previous incumbent. The payoffs are summarized in Appendix 1.
Patronage and Opposition Fragmentation
The problem is stationary in that all choice problems look the same at each election cycle, and for simplicity we assume no discounting of the future. Thus, the players’ optimal choices are the same each period (given the parameters and state variables), though outcomes can vary in each election because vote shares are stochastic. Let Vt denote the expected payoff to an unconstrained winner from election cycle t forward. This can be represented recursively as Vt = R – Dt (qt ) + p t+1(F t+1(qt ,Nt ))V t+1.
Consider the opposition candidates’ beliefs concerning how the cabinet vacancy will be filled at the end of an election cycle. Let πt (q t-1,N t-1,Nt ,Ft ) denote each opposition candidate’s subjective probability that the winner in election t will appoint a person outside the party (i.e., qt = 1). Opposition candidates’ expectations are rational, and in equilibrium subjective probabilities must equal actual probabilities, so πt = Prob(qt = 1|q t-1,N t-1,F t-1,Nt ,Ft ). Opposition candidates can influence πt via Ft , which reflects their joint decisions.
If a newcomer won the preceding election and filled the cabinet appointment from inside the party (i.e., N t-1 = 1 and q t-1 = 0), then what can be inferred by the opposition candidates regarding how the election-t cabinet appointment will be made in the event of an incumbent victory (i.e., Nt = 0) depends on the circumstances under which the election-t incumbent came into power as a newcomer in cycle t-1 (i.e., it depends on the value of F t-1). In particular, if the winner of the t-1 election was the leader of a unified opposition party (i.e., F t-1 = 0), then that winner was contractually obligated to appoint the opposition member (someone from his own party) to the cabinet office in election cycle t-1. So, if q t-1 = 0, N t-1 = 1, and F t-1 = 0, the opposition candidates recognize that q t-1 = 0 is contractually obligated and potentially a poor predictor of what the incumbent would do as an (unconstrained) winner of election cycle t. If, on the other hand, the winner of the t-1 election was a candidate in a fragmented opposition party (i.e., F t-1 = 1), then it can be inferred that q t-1 = qt , because the winner was unconstrained in election t-1 as well as in election t and faced the same problem in both elections. The following proposition describes our first result. All proofs are in Appendix 2.
Proposition 1: Opposition fragmentation in the current election is more likely to occur if the incumbent chose an outside appointment (i.e., opposition patronage strategy) rather than an inside appointment after the previous election.
To establish the result that opposition fragmentation in election t is more likely to occur when q t-1 = 1 than when q t-1 = 0, recall that fragmentation occurs as long as at least one opposition candidate plays “fragment.” Observe that “fragment” is a weakly dominated strategy for the opposition leader. Under unification, the opposition leader has a higher probability of winning the election than under fragmentation. Even if she loses under unification, she would receive the cabinet appointment with certainty if the incumbent chooses to coopt. The opposition leader has no incentive to deviate to “fragment.”
Our attention therefore focuses on the opposition member. If the opposition member observes q t-1 = 1, she infers that the incumbent (having appointed an outside candidate in the past) would appoint an outside candidate again in the event of an incumbent victory. This means that, under fragmentation, if the member loses, she would have a higher probability of being appointed to the cabinet post (conditional on an incumbent victory) than if q t-1 = 0. The member faces the trade-off that choosing fragmentation yields a greater (i.e., positive as opposed to zero) probability of victory at the cost of a lower probability of getting appointed to a cabinet post in the event of a loss. But this cost is smaller in the event of q t-1 = 1 than in the event of q t-1 = 0, making the member more inclined to choose “fragment.”
Equilibrium Analysis
Attention is restricted to subgame perfect Nash equilibria (SPNE) in pure strategies. Beliefs in equilibrium take a simple form. That is, conditional on q t-1, N t-1, F t-1, Nt , and Ft , opposition candidates can infer with certainty how the election-t winner will fill the cabinet vacancy; that is, πt is always either zero or one.Footnote 7 For technical reasons, we assume that the incumbent plays qt = 0 whenever the opposition member plays off the equilibrium path in election cycle t when deciding whether to fragment.Footnote 8
The equilibrium in the one-shot game is instructive and provides a useful benchmark. In that equilibrium, the election winner has no incentive to share power in period 3, so an appointment outside the party will never be made. The opposition leader anticipates this outcome and understands that whether the incumbent wins, or whether the opposition member wins (in the event of fragmentation), there is no chance that the opposition leader will ever be granted the cabinet position. With the cabinet position (and its payoff of u) off the table, the opposition leader’s only chance at a positive payoff is to win the election, and that happens with higher probability under unification than under fragmentation. The opposition member plays “fragment” only when p(0) is sufficiently high. Intuitively, when p(0) is high, the incumbent is likely to win, giving the opposition member a payoff of zero under unification. Thus, playing “fragment” gives the opposition member the best shot at a positive payoff. The bottom line is that the players know that patronage never occurs in the one-shot game with finite interactions among the players.
The situation changes in the case of infinite interactions because in that environment an incumbent reaps a future reward from sharing power in the present. That future reward takes the form of inducing opposition fragmentation in future election cycles, which in turn increases the incumbent’s likelihood of retaining power. The following proposition states the main result.Footnote 9
Proposition 2: (i) [Fragmentation Equilibrium] If the incumbent is weak and
$ \frac{d}{u} $
is sufficiently small, then an equilibrium exists in which the incumbent always pursues a strategy of opposition cooptation and the opposition always fragments.
(ii) [Unification Equilibrium] If the incumbent is strong and
$ \frac{d}{u} $
is sufficiently small, then an equilibrium exists in which the incumbent never chooses to coopt the opposition and the opposition always unifies.
Proposition 2 identifies two SPNE. The fragmentation equilibrium involves the opposition fragmenting and the election winner appointing outside the party, whereas the unification equilibrium involves the opposition unifying and the election winner appointing from inside the party.Footnote 10 Weak incumbents are incentivized to rely on the cooptation strategy, as doing so increases their chance of winning the next election. This, in turn, induces opposition fragmentation, as past outside appointments incentivize the opposition member to defect from forming a coalition. Already strong incumbents, on the other hand, will not pursue the cooptation strategy because the cost of making an outside appointment does not outweigh the benefit of a higher victory probability.
Data and Measurement
We corroborate the cooptation-based explanation for opposition fragmentation by empirically examining how politicians respond in electoral terms to an incumbent’s history of cabinet appointments. The formal model suggests that opposition politicians infer their likelihood of cooptation into government based on whether the incumbent has previously appointed opposition politicians to the cabinet. If that is the case, opposition politicians should be more likely to compete for office independently, splintering off from established parties or running on minor party tickets, rather than coalescing into larger parties. This line of reasoning implies a straightforward and testable hypothesis in executive elections: prior opposition appointments to the cabinet should be associated with a larger number of candidates vying for office in subsequent elections.
To assess this theoretical expectation, we develop original indicators of opposition cooptation via cabinet appointments made in African countries, most of which experienced political liberalization in the late 1980s and early 1990s, culminating in the introduction of multiparty elections. The region exhibits considerable variation in levels of democracy and incumbent turnover. We examine presidential elections held across 35 African countries between 1990 and 2016.Footnote 11 The sample includes virtually all presidential elections held in African countries during this period.Footnote 12
We find that the frequency of opposition cooptation is unrelated to the institutional constraints that African executives themselves face. African presidents do not operate under Westminster-style parliamentarism, so their mandates are independent of the legislature. Their ability to remain head of government is guaranteed by a fixed term and does not depend on maintaining the confidence of the majority in the legislature. Most legislatures either play no role in confirming cabinet appointments or have little influence over the president’s appointments to the cabinet. Reelected incumbents who go on to appoint opposition members to their cabinets, nevertheless, often do so despite winning a 61% vote share, on average.
Dependent Variable
The dependent variable is measured as the number of candidates who compete in a presidential election. The patronage-based argument implies that a larger number of candidates should be induced to contest an election if opposition politicians have been appointed to the outgoing government’s cabinet. Data on presidential candidate participation was compiled from print sources such as Nohlen, Krennerich, and Thibaut (Reference Nohlen, Krennerich and Thibaut1999), online sources like the African Elections Database, and news articles sourced from AllAfrica.com. The average number of presidential candidates is 8.6 and ranges from a minimum of 1 to a maximum of 33. As a robustness check, we also code an alternate version of the dependent variable that only includes candidates that received at least 0.5% or 1% of the total vote.
Independent Variable
We measure our main independent variable, opposition cooptation, by examining the appointment of opposition members to the cabinet. Politicians are identified as opposition members if they are not in the president’s party and they are members of a party that previously competed in elections against the incumbent. Given the frequency with which politicians move between parties as well as the speed with which they forge and dissolve alliances, we pay particular attention to the electoral history of the parties brought into the cabinet. We focus on whether a party has fielded candidates to compete against the incumbent in presidential elections or against incumbent’s party in legislative or local elections. This identification is consistent with our model, which is premised on the role of cabinet appointments made from outside the incumbent’s own party.
Identifying opposition cabinet members is straightforward in most cases. For example, in the year leading up to the 2010 presidential elections in Togo, President Faure Gnassingbé appointed two members of the Democratic Convention of African Peoples (CDPA) to his cabinet. In prior presidential elections, the CDPA had either competed directly against Gnassingbé’s Rally of the Togolese People (RPT) or supported other opposition presidential candidates. In 2010, they fielded their own presidential candidate. The CDPA representatives in the cabinet are therefore coded as being opposition members. In cases where a party’s participation in presidential elections is less evident, we examine their broader electoral history. In Nigeria, for example, in 2010 President Goodluck Jonathan appointed a cabinet minister from the All Progressives Grand Alliance (APGA), a regional party that had never previously run a presidential candidate against the ruling People’s Democratic Party (PDP). However, because the APGA did have a history of competing directly against the PDP for legislative and state-level elections, the APGA’s representative in the cabinet is coded as being an opposition member.
We only consider appointments to full-ranking cabinet positions when coding opposition cooptation.Footnote 13 We code instances of opposition cooptation using yearly data on cabinet appointments in all African countries. Data on individual opposition members appointed to the cabinet were compiled from annual volumes of Africa South of the Sahara and the Political Handbook of the World, corroborated through media searches in AllAfrica.com, and cross-checked using the WhoGov dataset (Nyrup and Bramwell Reference Nyrup and Bramwell2020).
To consistently identify opposition cooptation in the yearly cabinet data, we create incumbent spells defined as sequences of consecutive years for which an incumbent was in office and had an identifiable cabinet leading up to an election. Each incumbent spell ends in the year prior to the election.Footnote 14 Consistent with existing scholarship, we require that an incumbent spell last at least three years to be included in the dataset. If there is a nonpeaceful transition of power that interrupts an incumbency spell, such as a coup, we exclude such observations because they do not lead up to an election year. Relatedly, elections without a clear incumbent are excluded from the sample. Such instances can arise when a temporary or caretaker government is in place after a coup has removed the incumbent.
We employ two versions of opposition cooptation. First, we create a dichotomous variable, Opposition participation in prior cabinet, indicating whether any opposition members were appointed during the incumbent spell. This variable equals 1 if at least one identifiable opposition member was appointed to a cabinet position and 0 otherwise. We find that cooptation occurs regularly among African countries: 62% of incumbents in the sample appointed an opposition candidate to their cabinet sometime during their incumbent spell.
A second version of the independent variable, Opposition proportion in prior cabinet, is defined as the proportion of available cabinet seats appointed to opposition members. It is measured by dividing the total number of opposition appointments during an incumbent spell by the total number of cabinet seats available in that incumbent spell. Across African countries, the average proportion of opposition-held cabinet seats has oscillated over time between a low of 0.16 in 1990 and a high of 0.43 in 2009. The sample mean is 0.23, ranging from a minimum of 0 to a maximum of 0.92.
Control Variables
We control for variables relevant to party institutionalization, regime dynamics, and country conditions. We first account for differences in the nature of the prevailing party system. Incumbents may be constrained in their ability to make opposition appointments to the cabinet when their hands are tied by well-established parties that insist on reserving cabinet seats for their own members. We control for differences in levels of party system institutionalization (Party institutionalization) using an index from the Varieties of Democracy (V-Dem) Project designed to reflect the extent to which political parties are organized, ideologically coherent, and socially rooted (Coppedge et al. Reference Coppedge, Gerring, Knutsen, Lindberg, Teorell, Altman and Bernhard2020).Footnote 15 Since one of the most reliable indicators of party strength is whether the incumbent party is able to survive in power beyond the death or departure of its founding leader (Meng Reference Meng2021), we also use a dummy variable (Incumbent party survives founder) equaling 1 if the incumbent’s party has survived in power beyond its founding leader, and 0 otherwise. Additionally, opposition cooptation may be constrained by the existing size of the incumbent’s ruling coalition, so we control for the proportion of the legislature under its control (Incumbent share of legislature). This measure is calculated from the Database of Political Institutions (Cruz, Keefer, and Scartascini Reference Cruz, Keefer and Scartascini2018) by dividing the number of legislative seats held by government by the total number of legislative seats.Footnote 16
Following insights from the party systems literature on the importance of accounting for the interaction of electoral rules and sociological cleavages (Golder Reference Golder2006), we include a dichotomous variable for countries that require a second-round election if no candidate wins a majority of votes in the first round (Runoff system),Footnote 17 an index of politically relevant ethnic groups (Ethnic fractionalization) from Posner (Reference Posner2004), and the interaction between these two variables. Runoff systems are thought to weaken politicians’ incentives to coalesce, so they are expected to be associated with a larger number of presidential candidates (Cox Reference Cox1997). Greater ethnic fractionalization is expected to be associated with a larger number of presidential candidates (Brambor, Clark, and Golder Reference Brambor, Clark and Golder2006).
We also control for country-level political characteristics. Because political liberalization has yet to be consolidated in many African countries, we control for electoral experience and political rights. We use the V-Dem count for the number of presidential elections held since the last unconstitutional change of government (Consecutive presidential elections)Footnote 18 and the V-Dem measure for the extent to which liberal democracy is constitutionally achieved (Liberal democracy).Footnote 19 For robustness, we also substitute this variable with the V-Dem measure of electoral democracy and the Freedom House score for political rights.
To account for economic and sociological factors, we add a measure from the World Bank’s World Development Indicators for GDP per capita in constant 2010 US dollars (GDP per capita) as a proxy for the level of economic development. We also draw on the World Development Indicators to control for a country’s total population (Population) and oil rents as a percentage of GDP (Oil). We use the natural logarithm of these variables. See Appendix 3 for summary statistics.
Empirical Analysis
The statistical models use ordinary least squares (OLS) regression of pooled election data with country-clustered robust standard errors. This empirical strategy is consistent with prior studies of election cycles that conventionally use pooled OLS analyses (Elgie et al. Reference Elgie, Bucur, Dolez and Laurent2014; Golder Reference Golder2006; Hicken and Stoll Reference Hicken and Stoll2011). Well-known methodological problems arise when analyzing time-series cross-sectional data, including unobserved heterogeneity across units and serial correlation of errors, which may result in incorrect or misleading estimates (Beck and Katz Reference Beck and Katz1995). A fixed-effects specification can mitigate such problems under certain conditions but can also produce unreliable results, particularly when there are relatively few observations per unit and independent variables change only gradually over time, if at all (Clark and Linzer Reference Clark and Linzer2015).
In selecting an appropriate estimation strategy, we compared several potential approaches, including fixed effects, random effects, and detrended or demeaned data.Footnote 20 All of these alternative approaches show that the main explanatory variable of interest, opposition proportion in the prior cabinet, consistently attains statistical significance at the 0.05 level or better when examined alongside a core set of institutional and regime variables.Footnote 21 The results reported here are from the pooled OLS analyses.Footnote 22 However, because other studies of election cycles have used random-effects models (Lago and Torcal Reference Lago and Torcal2020; Weghorst and Bernhard Reference Weghorst and Bernhard2014), we also replicate our main results using random effects.Footnote 23
In further sensitivity analyses, we implement the test developed by Oster (Reference Oster2017) to gauge the extent to which omitted variable bias might affect our inferences. Specifically, we estimate the level of unobserved confounding that would be necessary to mistakenly detect a significant effect when the true effect is zero. For a regression with an R 2 of 0.4, which is consistent with our baseline analyses, we find that observations would have to select into cooptation on the basis of unobserved characteristics at a rate that is 6.7 times more than all the observable characteristics in order for the estimated effect of cooptation to disappear.Footnote 24
Past Opposition Cooptation Induces Opposition Fragmentation
Table 1’s results support the hypothesis that opposition appointments to the cabinet can be used to induce fragmentation. The coefficient on opposition cooptation in the prior cabinet attains its expected positive sign and is statistically significant in all model specifications at conventional levels. The estimated effects of opposition participation in the prior cabinet remain consistent regardless of which controls are included. Model 2 adds the measures for party system institutionalization, ruling party strength, and the ruling party’s share of the legislature, and Model 3 further adds the remaining institutional, sociological, and economic controls. The estimated coefficient on opposition participation in Model 3 retains its substantive and statistical significance, indicating that prior opposition cooptation is associated with nearly two additional presidential candidates in the subsequent election.
Table 1. Past Opposition Cooptation Increases Number of Presidential Candidates
Note: Pooled OLS regression. Dependent variable is the total number of presidential candidates. Robust standard errors clustered by country in parentheses. *** p < 0.01, ** p < 0.05, *p < 0.10, two-tailed tests.
Table 1’s Models 4–6 report the results when the main explanatory variable is switched to the proportion of cabinet seats allocated to the opposition in the years preceding the election. This variable’s coefficient is consistently positive and statistically significant with a large magnitude. Based on the associated coefficient in Model 6, increasing the opposition proportion in the cabinet by one standard deviation (0.26) is associated with 1.5 additional presidential candidates in the next election. Moreover, as the differences in R 2 across models suggests, the proportion of opposition members in the cabinet may account for more of the variation in the number of presidential candidates than the standard regime and institutional variables commonly employed in the study of election cycles.
Incumbents belonging to stronger ruling parties are more constrained in their ability to offer cabinet seats to the opposition, as reflected in the negative coefficient on party institutionalization. But the opposition cooptation results remain robust to controlling for ruling party strength and the ruling party’s share of seats in the legislature, as shown in Models 5 and 6. The coefficients for runoff system, ethnic fractionalization, and their interaction are statistically insignificant.Footnote 25 The V-Dem liberal democracy index suggests that the number of presidential candidates increases with greater democracy, but there is no effect associated with the number of consecutive presidential elections.Footnote 26
Our measure of opposition cooptation in Models 4–6 implicitly accounts for differences in resources by measuring opposition appointments as a share of total available cabinet seats rather than the total count of opposition appointments.Footnote 27 Nevertheless, we consider the fact that countries can vary in resource endowments that may facilitate the creation of extra cabinet seats that can be offered to the opposition. Model 6 suggests that the number of presidential candidates is correlated with level of development, but the coefficient on oil rents is statistically indistinguishable from zero.
Additional robustness checks presented in supplementary appendices show that the main results remain unchanged when accounting for a variety of other historical or electoral factors, such as controlling for boycotted elections (Appendix 7) or dropping the first election in the post-Cold War period (Appendix 11). Including the number of presidential candidates in the previous election, for example, has no substantive effect (Appendix 12). We also repeat the analysis using alternate measures of the dependent variable that only include candidates who received at least 0.5% or 1% of the total vote. These alternate measures roughly halve the maximum number of presidential candidates from 33 to 15, but the results remain consistent (Appendix 13).
We examine whether the estimated effects of cooptation depend on the use of other tactics commonly used to neutralize the opposition—namely, electoral fraud and electoral violence.Footnote 28 We find no systematic relationship between opposition cabinet appointments and the presence of fraud or violence in elections (the correlation is effectively zero), suggesting that cooptation is not a clear-cut substitute for illicit strategies of political control. The main cooptation results also remain consistent when we repeat our analysis while controlling for fraud and violence (Appendix 14).
We consider whether the changing nature of elections may condition our findings, particularly since opposition candidates can leverage the recent expansion of cellphone access to lower campaign costs (Bleck and van de Walle Reference Bleck and van de Walle2018). We therefore rerun the analysis to include a battery of campaign and party-related controls. These controls account for campaign finance rules,Footnote 29 restrictions on candidate or party entry,Footnote 30 restrictions on national broadcast media,Footnote 31 and the percentage of the population with access to internet to account for the use of social media in campaigning.Footnote 32 The results remain robust to the inclusion of these additional controls (Appendix 15).
Accounting for Alternative Hypotheses
While the models presented in Table 1 control for various aspects of party institutionalization, there may be concern that the results merely reflect underlying differences in the strength of opposition parties. Weak opposition parties could fragment even in the absence of the incumbent’s cooptation strategy. To account for this possibility, we repeat the analysis on a subset of elections that include only regimes with low levels of party system institutionalization (Appendix 16). The results hold even when we limit the sample to elections in which party system institutionalization falls below the mean on V-Dem’s index. The estimated coefficients on the opposition proportion of the prior cabinet are either larger or virtually the same as those in the main results.
Another concern stems from the nature of electoral rules. The cooptation logic is straightforward in plurality systems because incumbents obviously benefit from opposition fragmentation when they only need to win the largest number of votes. But does the cooptation logic equally apply in runoff systems? We claim that it does because incumbents are more likely to lose power if they must compete in a second round. Incumbents want to avert the very scenario that a second-round election manufactures—namely, the selection of a focal opposition candidate around whom others can rally. Incumbent incentives are borne out in the data: while 71% of elections in the sample were conducted in runoff systems, only 27% of those elections go to a second round. And whereas the opposition wins only 7% of first-round elections in runoff systems, they win 54% of second-round elections. Incumbents can thus anticipate that they are more likely to lose if the opposition is not sufficiently fragmented in the first round. Notably, incumbents who win reelection in the second round do so after facing a more fragmented opposition field in the first round.Footnote 33 We show that our findings remain consistent when limiting the sample to runoff systems (Appendix 17), excluding cases that go to a second round (Appendix 18), or limiting the sample to runoff systems with no second round (Appendix 19).
The cooptation strategy yields electoral benefits to incumbents even when they compete under the runoff system. Appendix 20 shows that the number of presidential candidates is positively and significantly correlated with the likelihood of incumbent election victory in runoff systems. Appendix 21 further shows that both the number of presidential candidates and the proportion of opposition appointments are positively correlated with incumbent victory in the second round. These relationships retain their statistical significance despite the limited number of observations.
An alternative interpretation for our findings is that opposition cabinet appointments reflect powersharing rather than cooptation (Ariotti and Golder Reference Ariotti and Golder2018). Incumbents who can be removed by a vote of no confidence in the legislature have an incentive to form alliances with other parties to remain in power. Directly elected presidents who exercise a mandate independent of the legislature have an incentive to engage in powersharing when they lack a legislative majority. In this respect, the powersharing and cooptation logics are not mutually exclusive. Incumbents might make cabinet appointments to simultaneously assemble legislative coalitions and induce intraopposition divisions.
To confirm that our results are not being driven solely by powersharing incentives, we reexamine our findings in relation to the composition of the legislature. The vast majority of incumbents in our sample (85%) enjoy a legislative majority with the average incumbent controlling 67% of legislative seats. But in 15% of the sample, the incumbent’s party does not control a legislative majority. As a robustness check, we rerun our analysis separately on two subsamples based on legislative control (Appendix 22). The results remain consistent across the two subsamples, suggesting that incumbents can induce opposition fragmentation whether or not their party is in the majority.
Another consideration is that powersharing may be more common in African countries emerging from civil wars. Existing scholarship has highlighted the role of powersharing in civil-war settlements (Adekanye Reference Adekanye1998). One might wonder whether our findings are driven by opposition appointments in postconflict regimes based on such powersharing agreements. We find that leaders in postconflict regimes do appoint more opposition members to their cabinets (Appendix 23, Model 1). As robustness checks, we control for postconflict observations (Appendix 23, Model 2) and, alternatively, exclude all postconflict observations (Appendix 23, Model 3). Our main results remain consistent in all such specifications.
Although many African parties have relatively weak ideological commitments, there are important exceptions. Opposition candidates who rely on ideologically committed partisans may be more reluctant to accept a cabinet position from the incumbent (Buckles Reference Buckles2019). Such opposition parties emerged in former liberation regimes such as Angola and Mozambique. Participation in political conflict hardens partisan bonds, which may prevent opposition members from defecting (LeBas Reference LeBas2013; Levitsky and Way Reference Levitsky and Way2012). We find that former liberation regimes are less likely to appoint opposition members to their cabinets (Appendix 24, Model 1). Our results remain robust when controlling for liberation regimes (Appendix 24, Model 2) and excluding such cases (Appendix 24, Model 3).
Weaker Incumbents Appoint More Opposition to the Cabinet
We now turn to a second observable implication from the theoretical model: weaker incumbents appoint a larger proportion of opposition members to the cabinet. While the analysis from the previous section focused on how past cooptation behavior influenced the number of presidential candidates, this analysis examines how election results affect the leader’s subsequent cooptation decisions. We use the leader’s vote share as a proxy for incumbent strength. Leaders who receive a larger vote share are interpreted as beginning their term from a position of strength. In Table 2, the dependent variable is the proportion of opposition members appointed to the cabinet following the election.Footnote 34
Table 2. Weaker Incumbents Appoint More Opposition Members to Cabinet
Note: Pooled OLS regression. Dependent variable is the proportion of opposition members appointed to the cabinet. Robust standard errors clustered by country in parentheses. *** p < 0.01, ** p < 0.05, *p < 0.10, two-tailed tests.
The results in Table 2 corroborate the expectation regarding weaker incumbents. As their relative vote share decreases, reelected incumbents appear to increase the proportion of opposition appointments to the cabinet. The estimated coefficients on incumbent vote share move in the expected negative direction and attain statistical significance at the 0.05 level or better. Model 7 suggests that a strong incumbent who receives nearly 80% of the vote, which is about one standard deviation above the regional mean, is likely to allocate about 15% of cabinet seats to the opposition. By contrast, an incumbent who receives 60% of the vote, about the regional mean, will give about 25% of cabinet seats to the opposition. Models 8 and 9 show that the inclusion of the full set of controls does not change the results.
The results presented in Table 2 remain consistent when we limit the sample to elections without opposition boycotts (Appendix 25). The main results are also insensitive to using random effects, fixed effects, and de-trended version data (Appendix 26 and 27).
The Electoral Efficacy of Cooptation
Is the cooptation strategy effective at keeping incumbents in power? To answer this question, we focus on the subset of weak incumbents, as we show in the last set of regressions that weak incumbents are more likely to pursue the cooptation strategy. Strong incumbents are safe in their reelection bids, regardless of whether they coopt opposition politicians or not, so including strong incumbents in a regression estimating the effect of cooptation on election results would bias the estimate downward. Again, we proxy for incumbent strength using election results. We restrict our attention to incumbents whose first-round vote share fell below the sample mean (61%).
We find that weak incumbents who pursue the cooptation strategy are more likely to win elections compared with weak incumbents who do not pursue the cooptation strategy. In Appendix 28, we show that opposition fragmentation can help to secure the incumbent’s victory. As the number of presidential candidates increases, so does the likelihood of incumbent victory. Additionally, we show that weak incumbents who appoint opposition politicians to the cabinet are significantly more likely to win reelection compared with weak incumbents who do not.
Conclusion
The theoretical and empirical analyses presented here help explain why informal patronage politics continue to overshadow formal institutions across a range of electoral regimes. Our cooptation findings illuminate a specific mechanism by which incumbents can manipulate partisan competition. We specifically show that entrenched incumbents can contain the threat posed by multiparty elections by encouraging more, rather than fewer, of their rivals to compete against them.
Our study raises several questions that require further investigation concerning the nature of patronage-based cooptation. Incumbents with finite resources cannot afford to coopt all politicians willing to join their governments. A leader whose rule is based on the distribution of patronage most likely has already assembled an oversized electoral coalition, and further additions to that coalition may require the reallocation of resources, possibly away from other regime allies or even redirected from social programs. How incumbents manage the costs associated with this recurrent recalibration, without provoking a revolt among regime allies, is unknown. More importantly, the costs that such a cooptation strategy imposes on institutional development and economic growth remain to be established.
Data Availability Statement
Research documents and data that support the findings of this study, including supplementary appendices, are openly available at the American Political Science Review Dataverse: https://doi.org/10.7910/DVN/R8TUOQ.
Acknowledgments
We thank Jaimie Bleck, Danny Choi, Michael Gibilisco, Kathleen Klaus, Michael Miller, Rachel Beatty Riedl, four anonymous reviewers, and participants at the American Political Science Association Annual Meeting, Midwest Political Science Association Annual Meeting, African Studies Association Annual Meeting, Institute of Social Science at the University of Tokyo, and UC Berkeley African Politics Working Group for helpful comments and discussions.
Conflicts of Interest
The authors declare no ethical issues or conflicts of interest in this research.
Ethical Standards
The authors affirm this research did not involve human subjects.
Comments
No Comments have been published for this article.