Africa's Lame Ducks: Second-Term Presidents and the Rule of Law

Abstract The vast majority of sub-Saharan countries have adopted constitutional clauses establishing that elected presidents cannot serve more than two mandates. While an extensive literature has examined why African leaders comply with or else try to manipulate term limits, the policy implications of the latter remain unexplored. Existing studies of other world regions suggest that setting a maximum number of terms presidents can serve tends to make them ‘lame ducks’ during their final mandate. We reconsider this argument, and posit and demonstrate empirically that constitutional limits can actually induce positive effects on second-term presidents' actions compared to their first terms. More specifically, the absence of electoral pressures, the concern for their post-presidential future and legacy-building motivations may lead to improvements in the rule of law, especially regarding the functioning of the judiciary. This article represents the first empirical investigation of the performance of Africa's second-term lame-duck presidents.

The politics of presidential term limits in Africa as well as in other world regions has attracted growing scholarly attention (Baturo and Elgie 2019;Grauvogel and Heyl 2020;Heyl and Llanos 2022, to name a few recent collections).Much of this literature, however, focused almost exclusively on the role such executive constraints play with regard to political and democratization struggles, including, in particular, their frequent manipulation or attempted manipulation (Dulani 2011;Maboudi et al. 2021;McKie 2019;Posner and Young 2018;Reyntjens 2020;Tull and Simons 2017;Versteeg et al. 2020; see Suzuki 2019 for a notable exception).
The questions that have been largely dodged are how and to what extent the regular functioning of term limits shapes African presidents' behaviours, priorities and policy decisions.Do presidential term limits have implications that reach further than preventing national rulers from overstaying in office?How, if ever, do they affect officeholders' governing behaviour and choices?Limiting a leader's time in office may be only part of the story.
To start filling this gap, we build on the idea borrowed from studies on other world regions (Besley and Case 1995;Johnson and Crain 2004) that governing with the prospect of running for re-election can be very different from governing without such a prospect.Hence, a leader's performance during the first mandate may deviate significantly from his/her second mandate.However, we differ from previous studies in one important respect.While much of the existing literature tends to highlight a 'lame-duck' syndrome and to associate incumbents' nonrenewable terms with negative policy outcomes, especially in the fiscal sector, we argue that term limits can also have positive consequences on second-term presidents' behaviour.
We contend, in particular, that second-term lame-duck presidents face three specific incentives whose interplay can lead to improvements in the rule of law.Such incentives stem from: the absence of direct pressures to manipulate the rules in order to ease re-election; a personal interest in a more impartial and accessible judiciary apparatus that will help shelter former presidents once they become ordinary citizens and possibly political adversaries of the new rulers; and a legacybuilding and reputational motivation.For these reasons, we should observe an improvement in African presidents' rule-of-law performance during their final term in office compared to their first term.
This article offers the first empirical comparative analysis of the performance of Africa's term-limited presidents.An opening section offers a brief overview of the debate on the politics of term limits in Africa.The next section looks at the existing literature on other world regions to introduce the policymaking implications of term limits with regard to the conduct of first-as opposed to second-term presidents.We then develop our theoretical argument about how term limits can lead to rule-of-law improvements in Africa and, in a further section, we offer a systematic picture of term limits and their application (or lack thereof) across the sub-Saharan region.Finally, we present the findings of a statistical investigation covering the 1990-2020 period.The analysis confirms that the rule-of-law performance of African term-limited presidents improves during the second term, especially regarding the judiciary.

Time's up
African rulers have long held their position with no end in sight.In the postindependence politics of the 1960s through the 1980s, many of them hung on somewhat indefinitely (Carbone and Pellegata 2020).When an end to personal rule came, it was often unforeseen and abrupt, caused by either a military coup, or else natural or accidental death.Politics and policies in post-independence Africa were hardly ever affected by the clear prospect of an end to a ruler's spell in power.
With the 1990s transitions to multiparty politics came new or reformed constitutions, which in the vast majority of cases envisaged formal tenure limitations for African presidents (McKie 2017).The politics of term limits in Africa has attracted substantial scholarly attention during the past two decades.Research has followed two main directions.First, a debate has emerged about the desirability of such institutional constraints, which remains disputed.The arguments in favour of term limits see them as helpful tools in advancing democracy in a region whose vulnerability to rulers' overstaying in office has been widely demonstrated (Jackson and Rosberg 1982).By forcing leadership change, term limits put into effect the principle that nobody should stay in power for too long, a notion African citizens widely support (Dulani 2015;McKie and Carlson 2022).
Moreover, term limits regularize a succession mechanism and timeframe.In this regard, while they could merely produce succession between leaders belonging to the same party without hindering or interrupting the latter's political hegemony, in practice they make opposition victories more likely by reducing the so-called 'incumbency advantage' (Baturo 2022;Cheeseman 2010;Maltz 2007;Venice Commission 2018).Government alternation, in turn, can be instrumental both in democratic consolidation (Huntington 1991)or, short of it, in 'liberalizing electoral outcomes' (Howard and Roessler 2006)as well as in broadening support for the political system (Bratton 2004;Moehler and Lindberg 2009).
Limited tenure, however, can hardly be claimed to be a requisite of democracy.The simple fact that parliamentary regimes typically allow an unlimited stay in office for prime ministers, as long as they enjoy legislative support, is a powerful reminder.Moreover, opposition victories can result from electoral manipulation too and the democratic commitment of opposition leaders should not be naively taken for granted (Wahman 2014).The very arguments for term limits can be turned upside down, as limiting re-election arguably limits voters' choices too (Corrales and Penfold 2014) and risks dissolving actual and perceived political accountability (De Palma 1994).
A second research strand examines what African leaders do when they reach term limits and, more specifically, their decisions to comply with or else try to remove or bypass legal restrictions.Indeed, the introduction of limited tenure by no means guaranteed the institutional and democratic progress of African polities.Term limits have been a key target for Africa's contemporary would-be autocrats who tried, and quite often managed, to cling on to power (Reyntjens 2020).Scholars investigated the motives that could either elicit or discourage African leaders' attempts to manipulate term limits (Posner and Young 2018), the strategies adopted (Tull and Simons 2017) and the factors affecting the probability of success (Cassani 2020;Dulani 2011;Hartmann 2022; see also Baturo 2014;McKie 2019).
While the politics of term limits in Africa has been at the centre of a lively scholarly debate, the consequences of the functioning of such limits on African presidents' broader behaviour, priorities and policy decisions have remained largely uncharted territory.A recent influential work on the politics of term limits across the globe (Baturo and Elgie 2019), for example, delves into issues related to democratization but devotes only a conclusive chapter to the above issues (Suzuki 2019).By shaping rulers' time horizons, however, term limits can significantly affect the way they act in their first and renewable term as opposed to their second and final mandate, which in turn can have major implications not only for understanding African political processes but also for development trajectories on the continent (Carbone and Pellegata 2020).

The lame-duck syndrome
Besides supposedly preventing national rulers from overstaying in office, do presidential term limits also shape African presidents' behaviours, priorities and policy decisions?Reflections about the impact of setting a maximum number of mandates for elected power-holders start from the problem John Carey pointed out with regard to lawmakers' term limits: since 'legislators' primary motivation is re-election … where re-election is impossible … the electoral connection is severed ' (1998: 3-5).The reasoning fully applies to elections for the executive too.By severing the electoral connection, term limits can make presidents' behaviour significantly different during renewable and non-renewable mandates.
A prevailing view, in particular, is that presidents who cannot be elected again are essentially weaker than presidents facing another election.The expression 'lame duck' is said to have been originally used for Calvin Coolidge, who was in the US White House for the better part of the 1920s.It refers to outgoing politicians.Its belittling meaning employs the image of a hesitant and uncertain conduct to convey a sense of precariousness and powerlessness.In its political substance, the expression implies a limited room for manoeuvre leaders have in their last mandate, and, relatedly, their own reduced incentives for manoeuvring.
Everybodylegislators, technocrats, lobbyists, businessmen, ordinary peoplewill be fully aware of the approaching expiration date, after which the outgoing president will lose whatever control he/she exerted on policies, resources, appointments and the like.Any of the chief executive's political agendas or promises would now appear somewhat less plausible, if anything, due to time constraints.This will in turn hurt and reduce political supportincluding legislative supportfor ongoing or new initiatives, which will likely decline.The last phase of a presidency will thus be an uphill struggle, and one a departing president may not even care to embark on, or would not want to undertake with any extra burden.
Principal-agent models of politics, and particularly reputation models, have been used to formalize the argument (Canes-Wrone and Shotts 2004).The strength of the incentives that elections, including re-elections, generate for politicians and their implications for policymaking are well documented (Meltzer and Richard 1981), especially when democratic and non-democratic regimes are compared (Lake and Baum 2001, among several others).Accordingly, incumbents who can run for office again are somewhat concerned with meeting voters' interests and demands, in order to cultivate consensus in view of the next election.On the contrary, non-re-electable presidents may place less value on, and thus less effort in, reputation-building, as removing the prospect of re-election implies disconnecting or softening the accountability link between elected and voters.
In a pioneering empirical work, Timothy Besley andAnne Case (1995, 2003) show that, while US state governors tend to keep taxes and spending low during a first term, ostensibly to strengthen their reputation in view of re-election, the reduced pressures for fiscal restraint when they can no longer stand for re-election leads to tax and spending increases, especially for Democrats.Similarly, James Alt et al. (2011) exploit the existence of different term-limit rules across US states to show that the performance of governors eligible for re-election is more in line with voters' preferences than the performance of non-re-electable incumbents. 1  Joseph M. Johnson and W. Mark Crain (2004) extend this reasoning by looking at 48 democratic countries.They similarly observe that tenure limitations affect fiscal outcomes, particularly by raising fiscal volatility, spending and taxes during final mandates.In turn, Chiara Dalle Nogare and Roberto Ricciuti (2011) find that chief executives who are not term-limited invest more than term-limited ones in social services and welfare.
The economic policy impact of term limits has been studied with regard to local administrators too.Alain de Janvry et al. (2012) apply the distinction between first-termers and second-/last-termers to assess the impact of electoral incentives on the performance of Brazilian mayors in implementing a federal programme of cash transfers aimed at reducing school dropout rates.While the programme was essentially successful across the country, municipalities governed by first-term, re-electable mayors performed significantly better than those under non-re-electable ones.
From a partly different angle, some studies posit that, as outgoing leaders approach the moment they will leave office, they begin to think of their tomorrow.In general, leaders with shorter time horizons may be more inclined to secure personal wealth by predating public and private resources (Olson 1993).Accordingly, outgoing leaders may want to put aside material assets as well as giving specific favours that will pay off at a later stage, without having to care about the risk of being sanctioned by voters (Suzuki 2019).Claudio Ferraz and Frederico Finan (2011), for instance, find second-termers in Brazilian municipalities to be less inclined to restrain corruption, compared to first-termers who can still run for election.
To summarize, most existing studiesa majority of them focusing on advanced democraciesassociate incumbents' non-renewable terms with negative behavioural and policy outcomes.Term limits themselves, it appears, come with their own limits.Taking this further, Johnson and Crain (2004: 86) question why newly democratized countries in developing areas such as Africa would want to adopt such constitutional arrangements, given their economic costs.
The good side of being lame While much of the existing literature tends to highlight the lame-duck syndrome, we argue that term limits do not necessarily shape the behaviours, priorities and policy decisions of African leaders in a negative way; thus, they need not be counterproductive to the development of African countries.
In particular, we identify in the rule of law an aspect of governance that is both crucial for the continent's development, and one in which term limits could actually trigger virtuous rather than detrimental cycles.Admittedly, 'rule of law' is a contested concept (Møller and Skaaning 2014), broadly referring to the supremacy of the law and the principle that no onecrucially, including the governmentis legibus solutus, and requiring that laws are enforced in a transparent, independent, predictable, impartial and equal way (O'Donnell 2004).So defined, the quality of the rule of law in a country is affected by multiple factors, such as the functioning of the judiciary, the pervasiveness of corruption and the efficiency of the public administration (Coppedge et al. 2021).
We focus on the functioning of the judiciary and, notably, on its independence and accessibility as well as on the extent to which the government complies with its decisions.Through a more thorough examination of the literature, we single out three main interrelated reasons why African presidents may behave in ways that could strengthen these rule-of-law dimensions during their last term in office.We refer to these mechanisms as electoral restraint, judicial insurance and legacy-making.
First, when moving from an initial renewable mandate to a conclusive one, electoral incentives essentially die out.Leaders whose survival in office depends on winning a forthcoming national vote may be prepared to go a long distance in trumping the rule of law and interfering with the judiciary to their own advantage in order to ensure the most favourable conditions for them to retain power.This may include deepening clientelist practices to widen and consolidate support (Cheeseman 2010;Ginsburg et al. 2011;Maltz 2007), limiting opposition rights to skew competition in their favour (Levitsky and Way 2010) and repressing dissent through force (Ritter 2014).Many such practices are extensively documented by the literature on electoral manipulation and the so-called 'incumbency advantage' (Cheeseman 2010;Levitsky and Way 2010;Maltz 2007).
Presidents who know they will no longer run for office and are on their way out, on the contrary, are less likely to engage in similar behaviours.For instance, Toke Aidt and Julia Shvets (2012) find that US state lawmakers who can no longer stand for re-election prove less inclined to misuse public spending.In a rare analysis of an African country (Tanzania), moreover, Anne Kjaer notices that 'leaders in their second term … do not have to follow appointment strategies that secure their longterm tenure of office ' (2004: 408; see also Dionne 2011).More generally, un-re-electable presidents will be comparatively less concerned with actively shaping the surrounding administrative and political environment.Thus, whether purposely or not, they are more likely to abstain from hindering or damaging what respect for laws and legal institutions their country can enforce.
Second, term limits may also offer reasons for outgoing incumbents who will soon return to private life to govern with moderation and to reinforce state institutions (Baturo 2014).Africa is a continent where risks of assassination, incarceration and exile for former presidents have traditionally been high (Goldsmith 2001).Mauritania's president, Mohamed Ould Abdel Aziz, who, despite picking his own successor, was jailed in 2021, a mere two years after leaving office, is only a recent case in point.Risks are raised by the fact that 'open seat polls'that is, votes not contested by an incumbent chief executiveoften result in opposition victories (Baturo 2022;Cheeseman 2010), and that opposition leaders are not necessarily 'democratic' (Wahman 2014).
Hence, we expect second-term presidents to be willing to make sure that the situation 'out there' will be reasonably safe for them, and that the next powerholders, who might be political enemies, will not have a free hand should they want to turn against their predecessors.The approaching possibility of a government turnover (and, relatedly, of a reversal of fortunes) may thus incentivize incumbents to buttress legal norms and due process, to reduce the room for future arbitrariness and ensure more benign conditions for a post-presidential future.One way incumbent rulers can make such support real is by voluntarily tying their own hands (Hirschl 2004) and subjecting themselves to existing rules and institutions, thus contributing to better legitimize and strengthen them, in an effort not only to establish a pattern of behavioural conduct for their successors (Howell and Mayer 2005), but also to insure against hostile political rivals who might gain office.
Third, second-term presidents will be increasingly concerned with building and leaving a positive legacy.Genevieve Kehoe (2014), for instance, questions the notion that second-term presidents become unproductive policymakers.She argues that, freed from concerns about political survival, most of them actually refocus on leaving something behind when they walk out.She refers to this as a 'terminal logic behaviour', resulting in frequent recourse to 'presidential proactive powers', notably decrees that in recent US experience took the form of executive orders, proclamations and memoranda.Focusing on the last 100 days of a presidency, William Howell and Kenneth Mayer similarly point out that 'outgoing presidents have every reason to strike out on their own' with 'last-minute presidential actions ' (2005: 534).
The will to leave a positive legacy might lead second-term presidents towards linking their name to the adoption of specific signature initiatives or landmark policies (Kjaer and Therkildsen 2012), such as the building of major national infrastructures, the resolution of enduring domestic conflicts or humanitarian crises, the striking of international deals for peace or regional cooperation (Potter 2016), and the like.Yet they might also want to leave a lasting mark by associating their name with a time of overall progress for their country.Achieving a significant international reputation may similarly motivate second-term leaders, in view of gaining post-presidency positions in multilateral organizations or private foundations, accessing perks such as consultancy jobs or foreign funding (Fruhstorfer and Hudson 2022) and obtaining recognitions such as international prizes and awards.
We may thus expect them also to make an extra effort to improve governance by strengthening justice and the institutions dispensing it.During his second term, for example, Kenya's president Mwai Kibaki pushed his chief justice to reform the judiciary while increasing almost sixfold the budget for the sector: 'I think he actually appointed me to bring about change based on the role I had played in constitution making.He wanted an outsider ….There was this thing, after my appointment: I was always introduced [by the president] to foreign dignitaries as "the man who will change the judiciary".' 2 To summarize: we contend that Africa's second-term lame-duck presidents face three specific incentives whose interaction can lead to improvements in the rule of law.Such incentive-based mechanisms stem from: (1) the absence of direct pressures for the incumbent to manipulate the rules and interfere with the judiciary so as to ease re-election (electoral restraint); (2) a growing personal interest in strengthening the impartiality and accessibility of the justice system, which will help shelter former presidents once they become ordinary citizens and possibly political adversaries of the new rulers ( judicial insurance); and (3) building one's personal reputation by leaving behind some relevant achievements (legacy-making).
The interplay of these three mechanisms leads us to expect term limits to induce positive rule-of-law developments during second-term presidencies, compared to first-term ones.More specifically, we derive the following hypotheses.
Hypotheses: Moving from first renewable terms to second final terms (and thus approaching the time to leave office), Africa's term-limited presidents become: Hypothesis 1: more compliant with the judiciary; Hypothesis 2: less likely to interfere with the judiciary; Hypothesis 3: more likely to promote improvements in the accessibility of the judicial system.
No doubt, counter-arguments can be made against the reasoning behind our hypotheses.One may contend, for example, that the expectation of being succeeded by a close ally makes a more malleable rule of law an attractive scenario for postpresidential illicit gains.Yet we believe that, particularly in Africa's often politically volatile contexts, the prospect of exiting office is bound to raise significant worries about how future power-holders will actually behave, possibly driven by a concern to constrain their influential predecessors too.Even the likes of Angola's João Lourenço and Botswana's Mokgweetsi Masisi, who both supposedly stood for political continuity, clashed with those who had handed office to them.We consider this to be a pre-eminent concern for presidents approaching the end of their mandates.
Similarly, it may be claimed that assuming presidents want to build and leave a legacy is essentially based on speculation, if not wishful thinking.We contend, however, that so long as such reasoning very frequently plays a role not only in the rhetoric of presidents themselves -Namibia's Hage Geingob or Kenya's Uhuru Kenyatta are good cases in pointbut, most importantly, in external observers' recurrent interpretations of their actions (Amuta 2021;Nossiter 2020;Tantoh 2019), the argument we make is well rooted in existing insights.
A reverse-causality objection may also be raised, based on the reasonable suspicion that lame ducks are themselves the product of a country's existing rule of law.This is a serious concern that we will tackle directly in our empirical analysis.For now, we simply observe that, in recent years, a number of African presidents appeared to choose to step down in poor rule-of-law performing countries such as Liberia, Sierra Leone, Mauritania, Niger and Burundi (if certainly not without controversies in the latter case): arguably, theirs was hardly a response to a wellestablished and respected legal framework.
Who are (and who are not) Africa's lame ducks?We want to investigate empirically whether African presidents have a better rule-of-law performance when their mandate is no longer renewable, as opposed to when presidents face the prospect of re-election.To test our hypotheses, we collect data on 180 leaders who held office in sub-Saharan Africa's 49 countries between 1990 and 2020, using the Africa Leadership Change (ALC) dataset (Carbone and Pellegata 2020), and we single out the directly elected, term-limited presidents who completed all the legally allowed mandates and then duly stepped down.
Overall, 40 out of 49 sub-Saharan states have thus far adopted constitutional term limits for their heads of government.Most of these countries have presidential or semi-presidential governments.Botswana and South Africatwo of the six parliamentary systems in the regionare exceptions in that they apply term limits to their indirectly elected executive presidencies.The remaining countries include four non-electoral regimes (Eritrea, eSwatini, Somalia, South Sudan), three parliamentary systems (Ethiopia, Lesotho, Mauritius) and a semi-presidential system with a term-limited head of state in which the executive is controlled by the prime minister (Cape Verde), and Gambia.
Sub-Saharan term limits mostly prescribe that a president cannot serve more than two mandates overall.Botswana sets a maximum of ten years in office, in line with the parliamentary nature of its government and the five-year mandate of each legislature.The constitutions of Equatorial Guinea, Guinea-Bissau, Mozambique and São Tomé and Príncipe establish a two-term limit but explicitly envisage the possibility for a leader to run for further non-consecutive terms.Uncommon arrangements have been in place for several years in both the Seychelles (a three-term limit until 2017) and the Comoros (a four-year rotating presidency between the three main islands during 2001-2018).
Since 1990, 41 directly elected presidents from 30 sub-Saharan countries completed the legally allowed executive mandates.Some leaders reached term limits more than once -Sam Nujoma (Namibia) and Pierre Nkurunziza (Burundi), for instance, reached them twice, while Blaise Compaoré (Burkina Faso) did so three times.Overall, term limits were reached on 45 different occasions. 3 As anticipated, attempts to either remove or bypass term limits have been relatively common.Leaders stepped down in due time in 21 cases.In the remaining 24 cases, they attempted to hold on to power, in various ways (for a typology, Cassani 2020).A majority of would-be overstayers were successful, the most recent case being Alassane Ouattara in Côte d'Ivoire.However, the list of aspiring overstayers also includes a number of presidents who were ultimately prevented from retaining power due to a military intervention (as with Burkina Faso's Compaoré, Niger's Mamadou Tandja and most recently Guinea's Alpha Condé), civil society and parliament's resistance (Bakili Muluzi in Malawi, Frederick Chiluba in Zambia, Olusegun Obasanjo in Nigeria, Joseph Kabila in Congo-DRC) or a defeat at the polls (Senegal's Abdou Diouf and Abdoulaye Wade).
These raw data offer quite a disheartening picture of Africa's term-limit politics more often than not, elected presidents showed little respect for the constitutions of their countriesand may lead to questioning the point of investigating the implications of term limits, that is, whether and how a formal institutional arrangement that has thus far proved highly malleable shapes the behaviour of political actors that so frequently engaged in its manipulation.Figure 1 shows, however, that there is another part of the story.By tracing the cumulative record of term limits' respect, manipulation and failed manipulation throughout the last three decades, the graph draws our attention to a point that is rarely acknowledged.
Respect for term limits is becoming increasingly frequent in the continent, especially since the mid-2010s.Since 2017, for instance, presidents stepped down voluntarily in five out of eight cases in which term limits expired, and in two of the remaining three cases they were eventually forced out of office. 4These trends suggest the time is ripe for exploring the implications of the regular functioning of these constitutional provisionsin terms of rule of law, as we do, as well as in other policy sectors.
Table 1 groups Africa's term-limited rulers according to whether they abided by constitutional norms, overstayed in office, or tried to do so but failed.'Law-abiders' represent those African presidents who actually became lame ducks during their second and final terms, and whose rule-of-law performance is examined in the next section.
Even a cursory look at the first column of Table 1 shows that compliance with term limits covers a rather heterogeneous group of countries and political leaders.In some countries respect for term limits is becoming routine, whereas Liberia and Mauritania have only recently first confronted term-limit expiration.Among the former group, respect for term limits is consolidating in countries quite different from each other from a political viewpoint: from Ghana, Sierra Leone and São Tomé and Príncipéthree of the relatively few African democraciesto Mozambique and Tanzaniaelectoral autocracies that are de facto hegemonized by a long-ruling partyto Benin and Kenyatwo hybrid regimes.Moreover, presidents stepped down in countries that previously experienced term-limit violations, such as Namibia, Niger and Burundi, and in unstable countries such as Mali.
Besides term-limited presidentsbe they law-abiders, successful overstayers or failed onesduring the past three decades sub-Saharan Africa has been home to several other kinds of political leaders.These include 'unlimited' presidents, directly elected under a constitution without term limits, such as Yahya Jammeh in Gambia, but also Maaouya Ould Sid'Ahmed Taya in Mauritania, who was in power when term limits were not yet adopted, and Faure Gnassingbé in Togo, who became president when term limits were no longer in place.Moreover, there are 'parliamentary' leaders elected or otherwise entrusted by a country's national assembly, 5 as well as 'unelected' leaders.The latter encompass rulers of countries that never transitioned to full electoral politics, rulers who served before their countries introduced elections (e.g.Sani Abacha in Nigeria), as well as rulers who grabbed power by the gun in countries that had already adopted electoral politics (e.g.Azali Assoumani in the Comoros).African presidents and the rule of law: An empirical analysis As a reminder, we expect African presidents in their final term to become more compliant with the judiciary (H1), less likely to interfere with the judiciary (H2) and more likely to improve the accessibility of the judicial system (H3), than in their previous term.To test our hypotheses, we conduct time-series cross-sectional regressions, adopting country-years as our units of analysis .Hereafter, we detail how we build and employ our variables and we present the analysis and the findings.For clarity, the first two indicators, which reflect the relationship between the executive and the judiciary, refer to courts other than the constitutional (or high) court, which helps us avoid endogeneity concerns raised by the fact that our analysis focuses on presidents respecting constitutional term limits.In turn, the third indicator measures the extent to which citizens can bring cases before the courts without risks to their personal safety, receive a fair trial and seek redress when public authorities violate their rights.

Independent variable: varieties of African leaders
We create a series of leader identifiers representing our main independent variables.The two most important leader identifiers are First term and Lame duck dummy variables that identify and distinguish African term-limited presidents who served two full terms and then stepped down with no attempt to hang on to office (i.e.presidents classified as 'law-abiders' in Table 1).These indicators take value 1 when these presidents are serving the first and the second/final mandates, respectively.
A third key leader identifier singles out the remaining term-limited presidents listed in Table 1, that is, Africa's Overstayers, both successful and failed ones.Three other leader identifiers distinguish Unlimited presidents, Parliamentary leaders and Unelected leaders.A final, residual (and thus more heterogeneous) category of Incomplete presidencies includes leaders who, as of 2020, never reached the end of their first term, ruled for one term only, or started but did not complete their final term.Earlier terminations are mostly explained by coups d'état (e.g.Marc Ravalomanana in Madagascar), resignations (e.g.Charles Taylor in Liberia), natural deaths (e.g.Levy Mwanawasa in Zambia) and electoral defeats (e.g.Nicéphore Soglo in Benin).This residual group also comprises interim leaders who only held the chief executive office for short periods (e.g.Joyce Banda in Malawi) and heads of transitional governments (e.g.Catherine Samba Panza in the Central African Republic). 6 While we are mainly interested in term-limited law-abiders and in seizing differences (if any) between the rule-of-law performance of Lame duck and First term presidents, we conduct the analysis within the broader comparative context of all African leaders.The exclusion of the remaining types of leaders would imply a significant loss of variance and information.For instance, it is useful to analyse Overstayers separately, as term-limit manipulation has been a relatively frequent practice south of the Sahara, and one highly consequential regarding the rule of law, typically requiring an incumbent to interfere with the judiciary to change the rules and/or to exploit legal loopholes.In turn, we do not expect Unlimited presidents to modify their behaviour depending on whether they are in their first or subsequent mandates.Moreover, we consider Parliamentary mandates to be inherently distinct from terms in presidential systems, due to the different reach and source of the executive power.Finally, sub-Saharan Unelected power-holders have historically ruled with few institutional constraints, and for this reason they are better examined separately from elected leaders.

Control variables
We consider several control variables that may influence the relationship under examination.Following Jørgen Møller and Svend-Erik Skaaning (2014), we account for the economic determinants of the rule of law.Modernization theory, for instance, links economic development with public pressure for individual rights and fair treatment by public authorities.We measure economic development both as gross domestic product (GDP) per capita levels (logged) and as GDP annual growth rates.
According to the 'resource curse' theory, however, an economy heavily reliant on natural resource revenues, which we measure as a percentage of GDP, short circuits the connection between citizens and the state.Moreover, we include an indicator of development aid (official development assistance, ODA, as a percentage of gross national income, GNI) to control for the emphasis that international cooperation programmes place on the rule of law (Carothers 1998).Economic data are from the World Bank's World Development Indicators.
To reduce the risk of omitting factors influencing both the dependent and the main independent variables, we also consider several political features which may affect both a country's rule-of-law performance and the respect of term limits, and thus also determine the lame-duck status of an African leader.First, we use the V-Dem Polyarchy Index, as countries scoring high in electoral democracy tend also to have a stronger rule of law (Maravall and Przeworski 2003) and to respect term limits (Reyntjens 2020).Hegemonic parties too, however, might be interested in preserving term limits as an internal succession mechanism (Ezrow 2019).Hence, we measure the consecutive years the same party has remained in power and check whether this interferes with the relationships we want to highlight.
Finally, we control for executive corruption using V-Dem dataas corrupted incumbents may either try to escape term limits (Baturo 2014), pack the judiciary with loyalists to protect themselves from future investigations, or bothand we create three dichotomous and time-invariant indicators signalling former single-party regimes (Hartmann 2022), former British colonies and countries whose constitutions envisage the possibility of running for additional non-consecutive terms.

Analysis and findings
As a starting point, since this is the first comparative investigation into the rule-of-law implications of term limits south of the Sahara, we want to gain a sense of the broad picture.Accordingly, Figure 2 uses the comprehensive Rule of Law Index to summarize differences in the average rule-of-law performance of different types of African rulers.
The graph vividly shows a major rule-of-law divide between a group of unelected, unlimited and overstaying presidents and another group of term-limited presidents and parliamentary prime ministers.While the latter outperform any other types of leaders, Figure 2 confirms that moving from the first renewable term to the second and final term corresponds, on average, to an improvement in terms of rule of law.Prima facie, African presidents display a better rule-of-law performance when they become lame ducks, and a positive effect of term limits on the rule of law does seem to exist.
A closer look at the data reveals that second-term improvements were recorded under presidencies as diverse (in terms of level of democracy, colonial and authoritarian legacies, leader's background, and level of development) as those of, among others, Benin's Thomas Boni Yayi, Mozambique's Joachim Chissano, Kenya's Mway Kibaki, Sierra Leone's Ernest Bai Koroma and Ghana's Jerry Rawlings.Unsurprisingly, however, we also found deviant cases that do not conform to the abovementioned pattern, that is, countries where the rule of law actually worsened during a president's second termsuch as Mauritania under Mohamed Ould Abdel Aziz and Niger under Mahamadou Issoufouor else remained persistently low, as with Pierre Nkurunziza in Burundi.
As a second stepand a more cogent testwe conduct time-series crosssectional regression to determine whether the term-limit effect previously detected holds true even when a larger set of factors that are likely to influence a country's rule of law are considered.In this case, we focus on the three more specific Executive compliance with the judiciary, Judiciary independence and Access to justice Rule-of-law performance is estimated using the V-Dem Rule of Law Index, an interval scale which ranges from 0 to 1.We omit from the analysis the residual Incomplete leader category.
dependent variables.For each indicator, we run a series of regression models.In all the analyses, we compute robust standard errors and we include independent variables that are one-year lagged with respect to the dependent variable, as well as year fixed effects.
We start from a simple model testing only the leader identifiers, namely, Lame duck, Overstayer, Unlimited, Parliamentary, Unelected, plus the residual Incomplete.First term leaders represent the reference category omitted from the regression model.Subsequently, we add stepwise economic, political and time-invariant control variables.Next, we drop from the model specification the time-invariant variables and we run other models adding alternatively country and leader fixed effects to control for any country-specific and leader-specific characteristics that may influence the relationship under examination, respectively. 7 Finally, we conduct another robustness test to address endogeneity concerns.To be sure, our research design envisages several strategies to account for this potential bias.On one hand, our Lame duck and First term identifiers only apply to African term-limited presidents that served two full terms and no more (De Janvry et al. 2012), while also considering other types of political leaders separately (including overstaying presidents).On the other hand, as discussed, control variables (as well as leader fixed effects) have been selected to reduce the risk of omitting from the regression analysis key variables that may determine both a country's rule-of-law performance and African presidents' propensity to respect term limits and thus to become lame ducks.
As a further strategy, we use instrumental variables and a two-stage least squares estimator to reassess the term-limit effect.Valid instruments must be relevantthat is, good predictors of the allegedly endogenous independent variable (X)and they must influence the dependent variable (Y) only through their effect on X (net of any controls).
To select our instruments, we mainly rely on the 'incumbency advantage' theory, which has a relatively strong explanatory power in Africa's electoral politics (Cheeseman 2010).In short, the theory suggests that sitting presidents tend to be stronger candidates at the polls and that it is easier to be confirmed in office than to be elected for the first time.Building on this idea, we use the winner's vote share and the margin of majority relative to the second-ranked candidate in first-round presidential elections to predict if African term-limited presidents are serving their first or second mandate.Exploring the data confirms that these variables are relatively good predictors of lame-duck status.At the same time, we do not expect electoral performance to directly influence the quality of the rule of law in the subsequent years. 8As a further instrument we use a president's age.Given the use of election data, estimating the instrumental variable models de facto requires excluding Unelected rulers from the analysis.
Tables 2-4 present the findings.Each table reports the same above-described battery of tests for a different rule-of-law indicator -Executive compliance with judiciary, Judiciary independence, Access to justice, respectively.In line with our argument and expectations, the overall evidence shows that respect for the rule of law tends to be higher during a term-limited president's second mandate, compared to first mandates.More specifically, the government is more compliant with the judiciary's decisions (Table 2 -H1), the latter acts more independently (    -H2) and citizens enjoy better access to justice (Table 4 -H3) during last nonrenewable terms as opposed to first terms.The results appear statistically significant at the 95% confidence level or higher, and robust to the several replication tests that we run and the several strategies we used to address endogeneity concerns.In only two cases, we obtained a weaker statistical significance (90%), namely, Model 6 in Table 3 and Model 1 in Table 4.We did not find systematic and sufficiently strong (from a statistical viewpoint) evidence regarding the other categories of leader, with the exception of African overstayers, whose rule-of-law performance is unsurprisingly negative, especially regarding compliance with judiciary and judiciary independence.In turn, the regression analysis does not confirm the apparently superior rule-of-law performance of parliamentary prime ministers suggested by Figure 2. As for our control variables, we find democracy and, to a weaker degree, one-party legacies and a past as British colony to have positive effects.Moreover, both the level and the pace of economic development show some evidence of a positive association.Conversely, corruption negatively influences all the examined indicators.

Conclusion
In no world region are lame ducks a native species.Africa was no exception, as post-independent politics long did without any precise timeframe for leaders to rotate in office.Things changed in the 1990s with the decision to couple the return to multipartyism with the introduction of directly elected, time-bound presidential tenures.From then on, an increasing number of sub-Saharan countries experienced incumbent presidents holding office for second mandates due to be their last term.
Non-re-electability has potential implications that go beyond the original goal of avoiding immovable leaders and favouring executive rotation.It can alter some key underlying political dynamics, shape the behaviour of leaders (as well as of other players) and generate distinctive development outcomes.Despite the vast attention gained by executive term limits in the sub-Saharan region over recent years, and the existence of an international literature pointing to the specificity of non-renewable spells in office, whether lame-duck presidencies have something special in Africa too had never been explored.
In this article, we thus ventured into this new research avenue by examining the broader impact on the rule of law of constitutional provisions that only allow African presidents two mandates in office.Our evidence shows that second terms are indeed different from first terms when we look at progress made along several indicators regarding the judiciary, in particular.
The results of this first empirical investigation of the performance of Africa's second-term lame-duck presidents represent an innovative contribution.Most importantly, our research challenges the view that setting limits to candidates' re-electability risks breaking the electoral connection and thus impeding accountability, ultimately generating disappointing government performances.In the rule-of-law field, African presidents on their way out actually appear to be essentially better than presidents who are to run for office again.
At the same time, with regard to the mechanisms we theorized, one should acknowledge that, while legacy-making dynamics may conceivably feature in any political system, the electoral restraint and judicial insurance incentives are more likely to emerge where state institutions are not fully consolidatedas is the case in much of Africa.This would admittedly limit the generalizability of our empirical results, as they might likely apply to other developing areas, but less so to the more fully institutionalized polities of advanced economies.
Our investigation of the policy and development implications of Africa's presidential term limits represents a starting point, as many related questions still lie ahead.Is Africa entirely immune from the lame-duck syndrome previously detected in other world regions?Or do second terms prove detrimental in this continent too, when we shift attention to other policy areas?Are African overstayers who justified manipulation of term limits by claiming that development demands continuity ultimately right?And, when the exercise of power remains authoritarian but embedded in institutions rather than fully personalized (Meng 2020), how does a maximum number of mandates interact with other types of constraints in shaping behaviours and policies?Further research will help expand our understanding of these and related issues by examining how tenure restrictions affect different social, economic and governance outcomes in the region.

Notes
1 Smart and Sturm (2013) offer an alternative theoretical argument according to which a two-term limit system is nonetheless preferable to a single-or no term limit system, as the former triggers a combination of selection and truthfulness effects. 2 Willy Mutunga, former chief justice of Kenya (author interview, 22 October 2019).Cf.Gainer 2016. 3 Other presidents left office by resigning before the expected end of their mandates, including Seychelles' France-Albert René and James Michel, and South Africa's Thabo Mbeki and Jacob Zuma. 4 It is also true that during the same period the likes of Paul Biya, Denis Sassou-Nguesso, Ismail Omar Guelleh and Yoweri Museveni started their fourth and then fifth consecutive terms, having scrapped term limits several years ago already.5 Most of these leaders hold the title of prime minister, but the category also comprehends Botswana and South Africa's term-limited indirectly elected executive presidents.6 The Incomplete residual category also includes presidents who ruled in Comoros between 2001 and 2018.These leaders only served one term each, due to the rotating presidency principle between the three main islands.7 Concerning individual fixed effects, see for instance Ferraz andFinan (2011), Aidt andShvets (2012), De Janvry et al. (2012), Geys and Mause (2016), Dalle Nogare and Kauder (2017).8 Correlation between our measures of electoral performance and the selected rule of law indicators in subsequent years (t + 1, t + 2, etc.) is always below −0.12 and progressively weaker.

Figure 1 .
Figure 1.Term-Limit Compliance and Manipulation in Sub-Saharan Africa Notes: Each line tracks the cumulative number of episodes of term-limit manipulation, failed manipulation and compliance, respectively.Data are from the Africa Executive Term Limits dataset (Cassani 2021).
Dependent variable: measuring the rule of lawConcerning the dependent variable, we use the Rule of Law Index from the Varieties of Democracy dataset (V-Dem,Coppedge et al. 2021), and we disaggregate it to analyse separately the indicators that are most relevant to our argument and hypotheses, namely: Executive compliance with the judiciary, Judiciary independence and Access to justice.

Figure 2 .
Figure 2. African Political Leaders and the Rule of Law Notes: The vertical bars represent the average rule-of-law performance of different types of African political leaders.Rule-of-law performance is estimated using the V-Dem Rule of Law Index, an interval scale which ranges from 0 to 1.We omit from the analysis the residual Incomplete leader category.

Table 2 .
African Leaders and Executive Compliance with Judiciary (H1)Notes: *** p < 0.01, ** p < 0.05, * p < 0.10.All predictors lagged one year.Robust standard errors clustered by country in parentheses.Model 7 is based on instrumental variable regression (two-stage least square estimator).The dependent variable is measured using the model estimates version of the V-Dem indicator Compliance with judiciary, as suggested in the V-Dem Codebook.

Table 3 .
African Leaders and Judiciary Independence (H2) Notes: *** p < 0.01, ** p < 0.05, * p < 0.10.All predictors lagged one year.Robust standard errors clustered by country in parentheses.Model 7 is based on instrumental variable regression (two-stage least square estimator).The dependent variable is measured using the model estimates version of the V-Dem indicator Judiciary Independence, as suggested in the V-Dem Codebook.

Table 4 .
African Leaders and Access to Justice (H3) Notes: *** p < 0.01, ** p < 0.05, * p < 0.10.All predictors lagged one year.Robust standard errors clustered by country in parentheses.Model 7 is based on instrumental variable regression (two-stage least square estimator).The dependent variable is measured using the V-Dem index Access to Justice, which ranges from 0 to 1.