The Politics of Legislative Expansion in Africa

Abstract

The number of seats in national legislatures around the world rarely changes. Yet, in Africa, a substantial number of countries have regularly increased the size of their legislatures, and these increases have become more common in recent years. Previous research on political offices in Africa’s electoral autocracies has suggested that their numbers and increases are largely motivated by patronage and clientelist considerations. Is this also the case for national legislatures? Curiously, very little political science scholarship exists on legislature size, either in Africa or the rest of the world. Using a mixture of descriptive statistics to present a new database, as well as econometrics and three case studies, we find that legislative expansion can be linked to executive branch manipulation. Presidents have found it politically useful to expand the size of African legislatures to weaken and/or control it.

Keywords

African politics legislative studies

Introduction

The multiparty¹ electoral era that began in sub-Saharan Africa in the early 1990s has been characterized by the phenomena of legislative expansion.² Of course, as in other regions, national African legislatures vary significantly in size, at least in part because of differences in population size. Even adjusting for population, significant variation in the size of individual constituencies can be observed in Africa, as elsewhere. Where African political systems have been different is in their proclivity toward expansion, particularly since the early 1990s.

African legislative expansion takes two forms. First, in a number of countries, the number of seats in the legislature has gone through multiple changes. For instance, Angola’s legislature included a modest 53 members at its independence in 1975 and has today some 220 members, following four different increases in size. In Mauritania, for its part, the six parliamentary elections since 1990 have witnessed 5 seat increases, as the legislature has increased from 79 seats in 1992 to 157 seats in 2018. During this time as well, a standard two-round majority system has been replaced by a complicated mixed system, combining elements of both closed-list proportional representation and single-member two-round majority voting, with set-aside seats for women candidates. Overall, since independence, the size of the African legislature has been changed 230 times in some 45 African countries. Already present under the one-party era, this process of expansion has accelerated with the turn to multiparty electoral politics in the 1990s.

Second, several African countries have recently added a second house. Overall, some 24 countries (out of 49 in the region) currently have had some period of bicameralism in their postcolonial history or once had it (4 of these countries). Nonetheless, the predilection for bicameralism is relatively recent. It can be said that bicameralism is also an innovation of the contemporary multiparty electoral era, as 16 African countries have reestablished upper houses in the last 20 years (for an account of the recent turn to bicameralism, see Coakley, 2014).

What drives this proclivity toward expansion? The comparative institutionalist literature has mostly ignored the issue of legislative expansion. We turn instead to the literature on authoritarian resilience, which argues that executive control of the legislature constitutes a reliable mechanism of regime control, thus providing a motivation for executives to manipulate the legislature (Slovik, 2012; Schedler, 2013; Schuler & Malesky, 2014; Gandhi & Przeworski, 2006). Unfortunately, this literature remains quite abstract concerning the mechanisms by which control is asserted and does not specifically link the legislature’s size to executive manipulation.

In this paper, we focus on changes in the size of the legislature and argue that this proclivity toward expansion is engineered by the executive branch of government for self-serving political reasons.³ In particular, we link it to the relationship between the executive and legislative branches, notably as it is set by various formal institutions. Our initial hunch that there would be a straightforward correlation between legislative expansion and broad democracy measures such as those of Freedom House and V-DEM is only partly confirmed by our data. Clearly, legislative expansion occurs at different levels of democracy, though this may be because of a limited range of regimes in the region, none of which are consolidated democracies. Instead, we find that the decision to increase the number of seats in African legislatures originates in the executive branch of government and is usually opposed by the legislative opposition, some of the president’s backbenchers, and the independent media. More specifically, our data suggests that legislative expansion is likely to occur when the executive branch has the capacity to control the process. In addition to testing alternative hypotheses outlined by existing scholarship, we develop three overlapping hypotheses about the circumstances under which the executive branch finds it possible and convenient to resort to legislative expansion as an instrument to weaken or better control the legislature. Executives do so to weaken the House’s ability to function as a body, to more effectively control it, and/or to gain patronage positions for its own backbenchers and other party elites. We argue that the degree to which the president is then able to assert his or her preferences on the matter is less related to the level of civil and political rights in the country than it is to the ability to act on these three sets of motivations, which vary across our cases.

We use event history analysis and find suggestive supportive evidence for our hypotheses, with an original dataset of the legislative history of all of the countries in the region that have conducted legislative elections since 1990. Our results are robust to various specifications and models. The cross-national econometric analysis does not, however, leverage much certitude regarding causal mechanisms, and the paper turns to case studies of legislative expansion in Tanzania, Uganda, and Zambia to gain a better understanding of the processes of enlargement. The first two are examples of regular and regime-promoted legislative expansions. On the other hand, Zambia did not undergo a single increase from the beginning of the Third Republic in 1991 until a small increase in anticipation of the 2016 elections. This negative case provides additional insights into the reasons for expansion.

Together, the econometric and case study evidence strongly confirms that legislative expansion results from executive branch initiative, even if no single political logic can account for all episodes of expansion, and even if it is hard to distinguish the three motivations for expansion. The econometric analysis is suggestive, and the three cases tend to confirm that expansion can be explained with reference to clientelist and patronage considerations linked to the needs of the incumbent party and political control of key constituencies. In addition, however, expansion episodes also appear to frequently result from the political survival strategies of African presidents, who wish to ensure their control over a legislature in which the governing party shows signs of weakness or independence from the executive. In sum, in the contemporary era of electoral autocracies, we argue that efforts to increase the size of the legislature have been directed by the executive branch of government, almost invariably to either weaken the legislature’s effectiveness and/or to ensure presidential control over it.

Our findings have significant implications for our understanding of politics in contemporary electoral autocracies. First, they shed new light on the dynamics of legislative-executive relations in these regimes and the ways in which executive interference in the design of legislative institutions can be a mechanism to structure this relationship in order for the executive to escape the constraints of horizontal accountability. Second, our work builds on the seminal work of Opalo (2019, 2020) and shows that African legislatures are not epiphenomenal and institutional variations should be taken seriously. Our paper challenges a persistent theme of contemporary African politics that, despite the transition to multiparty electoral politics, formal institutions do not matter to political outcomes, at least compared to personal agency and informal institutions (for nuanced discussions, see Hyden, 2012; Van Cranenburgh, 2009; Tull & Simons, 2017). Yet, our findings about the willingness of incumbent executives to seek to alter the makeup of the legislature strongly suggest that they disagree and believe that formal political institutions can be instrumentalized to significant political advantage. If so, existing formal institutions should be understood to undergird executive power and, more generally, political equilibria in contemporary Africa in critical ways that have lasting consequences for these political regimes and, in particular, executive-legislative relations and the future prospects for democracy in the region.

The paper is divided into four sections after this introduction. The next section describes the evolution over time of African legislatures and compares them to those of other regions. The Theories of Legislative Expansion section then examines different plausible and testable explanations for legislative expansion. The Theory section then describes our empirical strategy for testing these explanations. The Empirical Strategy section presents the data and econometric results, while the Data section develops three brief case studies of legislative expansion to better understand their political dynamics. The Results section summarizes the findings and their implications before concluding.

African Legislatures

Because the size of western parliaments has been relatively stable over the last decades, we focus on regions that experienced the third wave of democratization. By comparing the data, we show that this proclivity toward legislative expansion is particularly salient in sub-Saharan Africa and thus warrants further investigation. Although Table 1 suggests that African legislatures are not the largest ones, they change size more often and tend to expand. Whereas Latin American and Eastern European parliaments have a similar number of changes, African legislatures changed size 142 times over the last 30 years.

Table 1.

Descriptive Statistics (1985–2019).

	Latin America (N = 23)	Eastern Europe (N = 28)	Sub-Saharan Africa (N = 44)
Average size of the legislature	192	232	174.4
Average size of the lower chamber	165.6	204.3	157.74
Number of changes in legislatures’ size	84	86	142
New upper houses	1	8	17
Number of increase	62	32	103
Number of decreases	22	54	27

Showing the evolution of the average number of seats by region and per year, Figure 1 shows two interesting features. First, we observe an interesting convergence which suggests that states took time to find a legislative equilibrium after the third wave of democratization. Recent variations in sub-Saharan Africa show that the continent’s parliaments are still in flux and that close attention should be paid to future developments. Second, Figure 1 provides further evidence that legislative expansion is significantly more prominent in sub-Saharan Africa. Following the collapse of the Soviet Union, Eastern European parliaments saw their size decrease dramatically. In Latin America, we observe a slight increase in the 1990s followed by a plateau at 200 seats. These two trends contrast with the sharp rise observed in sub-Saharan Africa over the last 30 years. With an average size of 150 seats at the beginning of the third wave, African parliaments now have more than 200 seats.

Figure 1.

Evolution of legislatures’ size (1990–2020).

If legislative expansion is more important in the African continent, we also observe a great deal of variation at the regional level. African legislatures vary in size, from 33 seats in Comoros, with its population of less than a million, to Ethiopia, with 655 members of the legislature, for a population of some 110 million citizens. The country’s size clearly has a strong predictive effect on the size of the legislature, but there is still a fair amount of variation in the size of individual constituencies. Thus, neighboring Zimbabwe and Zambia have relatively similar populations of around 17 million in 2019, yet a legislator in Zimbabwe represents some 45,000 citizens on average, compared to 102,000 in Zambia.

Overall, the general trend is a striking one of a great deal of volatility, with regular increases in size. If the Botswanan and Malian legislatures have only changed size twice since 1990, the Malagasy parliament’s size has been altered eight times. The average legislature in the region has increased 3.2 times.

Although one-party states underwent legislative expansion, the third wave of democratization has accentuated this trend. Over the last 30 years, 88.8% of African countries have added seats in their legislatures. Only 24 countries have undertaken decreases in the size of their legislature. 37.7% of the countries in the region have added an upper house since 1990.

Theories of Legislative Expansion

Legislative size has been largely neglected in the political science literature, and the few studies available fail to explain why this phenomenon is prevalent in sub-Saharan Africa. A rare study by Taagepera (1972; see also Stigler, 1976) links the size of the national legislature to population (for a formal treatment, see Shugart and Taagepera, 1989, pp. 173–183). In addition, Taagepera (1972) argues that the degree of “social mobilization” matters, which he measures by the percentage of the adult population which is literate. Along the same lines, more recently, Herman (2015, p. 8–10) links legislative size to population and to social and geographic factors, such as the need to ensure “adequate representation” to various economic, ethnic, linguistic, religious, and occupational groupings in society.

These accounts have three main limits. First, they do not offer systematic evidence for the relationship between legislative size and population growth. In fact, our data suggests that there is no correlation between the two variables. Second, neither Taagepera nor Herman link changes in the parliament to political factors external to the legislature itself. In fact, as we shall see, expansion is often driven by the executive branch, despite the opposition of many if not most members of the chamber. Finally, while these studies generally attempt to explain the cross-national variation in legislative size, none seeks to explain the actual proclivity for changing the size of legislatures.

On this topic, we could find only the work of Jacobs and Otjes (2014 and 2015). With a dataset comprised of 133 democracies between 1945 and 2008, they conduct an event history analysis that links the effective number of parties to increases in size, and economic recession to decreases in legislative size, thus advancing the more general argument that institutional innovations in electoral democracies are often promoted by political parties in pursuit of their own self-interest (See also Boix, 1999; Benoit, 2004; Colomer, 2005). In this case, small parties are more likely to pursue legislative expansion, Jacobs and Otjes argue, to increase their own representation in the legislature. The only regime variables included in their analysis are the age of the democracy and a dummy for bicameralism, and they find that young democracies and bicameralism are both associated with increases in the size of the legislature. In long-standing democracies, changes in the size of the legislature are typically decided by an independent and largely a-political body, such as the United Kingdom’s “Boundary Commissions,” or is determined by the legislature itself. In the United States, for instance, the House of Representatives has not increased in size since the 1929 Apportionment Act capped it at 435 members (Bartlett, 2014). Because most African states are electoral autocracies, such findings do not help us make sense of the proclivity toward expansion in the region.

Hence, we turned to other literature to understand legislative expansion. First, the recent literature on authoritarian resilience has theorized the legislature’s role in ensuring the survival of authoritarian regimes. As Svolik asserts, the role of the legislature in authoritarian regimes is to “enhance the stability of authoritarian power-sharing by alleviating commitment and monitoring problems” (Svolik, pp. 12–13; see also Brownlee, 2007; and Gandhi & Przeworski, 2006). Although these findings bring valuable insights into the executive-legislative relationship in autocratic settings, the mechanisms outlined remain either overly abstract or limited to informal mechanisms of intimidation, cooptation, and manipulation (for instance, Schedler, p. 64–5, 2013). Scholars such as Opalo (2020; see also Prempeh, 2008) emphasize the constitutional provisions that ensure the upper hand of the executive vis a vis the legislature, notably decree powers and the power to decide when the legislature can convene, but the literature has not assessed the impact of the size of the legislature on its independence from the executive. Clearly, autocrats have a substantial “menu of manipulation” in Schedler’s apt phrase that still needs to be properly cataloged and better understood; in this paper, we show that legislative expansion is often a favored strategy for authoritarian leaders in the African region.

Second, the recent literature on decentralization and administrative unit proliferation in sub-Saharan Africa provides interesting leads. Since 1990, the number of lower-tiered administrative units has increased by at least 10% in most African countries, thus leading to the creation of new constituencies (Hassan & Sheely, 2017). Although the international community originally incentivized African leaders to decentralize to foster development (Green, 2010), the literature has repeatedly shown that these expansions are politically motivated (Kimura, 2013; Hassan, 2016). Even in more democratic settings, such as Ghana, the executive branch has created new districts in order to help guarantee a majority in the legislature (Resnick, 2017). District creation can not only lead to more safe seats, but in addition, they can provide the party with useful resources at the grassroots level, which is important to mobilize voters and strengthen the party (Hassan & Sheely, 2017). Although not all legislative expansions are related to decentralization measures (e.g., creation of reserved seats), and district creation is not always linked to legislative expansion, these scholars’ contributions allow us to hypothesize that incumbents have been weaponizing legislative size for political motives, often linked to their attempts to project power outside of the capital.

Theory

Even the most cursory journalistic accounts of legislative expansion in African countries suggest that it does not result from party initiatives and is instead a highly political process that engages the very nature of the political regime. In Niger in 2014, the government proposed a bill in the National Assembly to increase the size of the legislature (from 113 to 117) during the next legislative election (2016). The government claimed such an increase was necessitated by the recent increases in population, which had created gaps in political representation.⁴ The justification for a legislative increase in Senegal in 2017 was quite different, though, once again, the impetus came from the government: the size of the parliament was increased by 15 in order to represent the sizeable diaspora.⁵ Other African governments (e.g., Rwanda and Tanzania) also used legislative expansion to improve women’s representation in parliament by adding reserved seats.⁶ In all cases, these governmental justifications suggest that one potential explanation for legislative expansion across the continent is the willingness to improve democratic representation and address malapportionment issues.

However, in all three cases, government justifications for expanding the size of the lower house were met with unambiguous hostility from the legislative opposition, members of civil society, and the media. In all three cases, the main argument against expansion was that it would favor the government and undermine democracy. In Niger, for instance, the civil society complained that the decision was unilateral and “cavalière.” Members of parliament brought a petition before the Conseil Constitutionnel to challenge the constitutionality of such measures, but the Conseil upheld the government’s measures.⁷

We argue that this proclivity toward expansion is engineered by the executive branch of government for self-serving political reasons, even if governments can plausibly argue that population increases or cultural diversity warrant the reform for representational reasons. Strategies such as legislative expansion have become particularly useful for incumbents operating within the formal constraints of competitive autocracies. Because of the nature of these regimes, horizontal accountability has become a point of contention between executives and not only the opposition but also, in many cases, the incumbent party’s own backbenchers who would also like to constrain presidential power. Strategies undermining check and balances mechanisms have thus become a helpful tool to ensure the stability of regimes. More complex and easily justifiable, these reforms allow executive branches to manipulate the rules of the games for their own advantage.

Adding seats to the legislature has three main political advantages to electoral autocrats. First, it stands to reason that a smaller parliament is likely to be more politically powerful, as the power of individual legislators will be greater, on the one hand, and collective action problems will be more easily resolved, and the legislature will be able to act more boldly. As Kreppel (2014, p. 122) suggests, “more members are likely to lead to more complex internal organization mechanisms, and more thinly spread institutional resources.”

Second, an increase in seats allows governments to expand their clientelistic networks. A number of scholars have posited that public offices in Africa have proliferated since independence for clientelist or patronage reasons (Arriola, 2009). Much evidence attests to this proliferation: the size of government cabinets has long tended to increase (Van de Walle, 2001), while the numbers of districts (Ayee, 2013; Green, 2010; Hassan & Sheeley, 2017; Resnick, 2017) and other subnational units have grown considerably (Grossman & Lewis, 2014), as have the number of generals in the army. In this case, legislative size might depend on the complex combination of factors that influence the degree of clientelism and elite patronage in African countries, from fiscal resources to governance dynamics. For instance, with autocrats facing varying levels of intra-party dissent (Opalo, 2020; Meng, 2020), adding seats to the legislature offers an opportunity to expand patronage control by providing a position and status to supporters.

Finally, legislative expansion increases the number of backbenchers supporting the government. Recent scholarship has shown how African legislatures have been critical for maintaining political stability (Opalo, 2019). Moving away from the patronage account, this literature seeks to explore how the executive branch has developed tools to control the legislature. In this case, the president’s objective is less to weaken the legislature as a decision-making body than to control it better. The government’s motivation to pursue expansion is the creation of safe seats, presumably either by malapportionment, gerrymandering, or specific electoral rules, quotas, or non-elected seats. The objective is not only to pad the presidential majority but also to increase the number of backbenchers who will more reliably support the president. The evidence for such practices is easily found in the Africa region. As we shall see, for instance, electoral autocracies in Rwanda, Uganda, and Tanzania have increased the size of their legislature in recent years to add women’s seats; indeed, they have been praised for their efforts to increase the political role of women. Although governments have relied on different electoral mechanisms to improve women’s representation in parliament, in each case, seats have been primarily filled by pro-government members (e.g., Muriaas & Wang, 2012). Careful accounts have shown the high level of legislative malapportionment resulting from expansions in countries like Cameroon (Albaugh, 2011, p. 121), Tanzania, and Uganda (see the case study material below).

By outlining these three different advantages, we aim to show that no one-size-fits-all explanation can account for all the variation in legislative expansion across the continent. Depending on the political context and other institutional factors, the rationale may vary. For instance, under certain institutional configurations, the cost of legislative expansion might outweigh its benefits. To understand variation in legislative expansion across the African continent, we must pay attention to additional institutional constraints beyond the legislative-executive relationship. First, we hypothesize that legislative expansion is less likely to occur in states governed by proportional representation as this electoral system introduces greater uncertainty for incumbents and would a priori not improve the position of the president’s party. Second, because an independent electoral commission might prevent the executive from allocating seats in constituencies that favor the ruling party, we suspect that the executive will not rely on legislative expansion when they do not control the electoral commission. This may notably be the case in the region’s most democratic states.

From our theory, we deduce three observable implications. First, legislatures are more likely to change size in states with a strong executive branch. To impose expansion, the executive branch must have both the power and the constitutional capacity to do so. Therefore, we expect to observe this phenomenon in less democratic states. Second, legislative expansion is more prone to occur in hybrid regimes where executives must nonetheless abide by democratic institutions to some extent. In more autocratic regimes, executives have the power to alter the legislature but fewer incentives to do so. It may remain attractive as a source of patronage but is not necessary to control an already pliant legislature.

Empirical strategy

We rely on a mixed-method approach to test our hypotheses. An original cross-sectional dataset of African legislatures is used to run an event history analysis to identify statistically significant associations between important political factors and build a robust argument about the causes of legislative expansion. Cross-sectional analyses have their limits, however. We cannot make any definitive causal claim or even confidently determine the direction of the causal arrow between our primary explanatory variable and our outcome of interest. To remedy these shortcomings, we process trace the politics of legislative expansion in three case studies. First, we seek to identify the actors responsible for the legislative expansion of the Ugandan and Tanzanian parliaments and understand their motivations. We then exploit a source of variation by looking at a counterfactual—the case of Zambia.

Among many parametric and semi-parametric models available, we use the Cox Duration Model for several methodological reasons. First, because the Cox Duration Model does not specify the distributional form of the hazard rate, it minimizes approximations on the form of the duration dependency (Jones & Branton, 2005). Second, this model is particularly useful in our case since it allows for a longitudinal analysis of factors that could affect “the survival of the legislature” (Jacobs & Otjes, 2015).

Our model complies with the different sets of assumptions of the Cox Duration Model. One major threat to our findings could be the lack of independence between the numerous changes in the size of the legislatures. For instance, one could make the case that parliaments which have experienced a change in size in the past are more likely to see their composition altered in the future. For this reason, we test for recurrent events using a shared frailties model (Appendix A) with a gamma distribution fit to account for such a possibility (Jacobs & Otjes, 2015; Mills, 2011). We also conducted a Schoenfeld residual results (Appendix B) and show that our estimates do not violate the proportional hazard assumption.

Data

We collected cross-sectional data on the size of African legislatures over a 30-year period (1990–2019). Although legislatures were not completely powerless before the third wave of democratization (Opalo, 2019), the absence of opposition made the legislature easier to manipulate by the single party in power. Following the wave of democratization that swept through Africa between 1989 and 1995, the relationship between the executive and the legislative branches changed as multiparty competition was generalized across the region (Bleck and Van de Walle, 2018). Today, even authoritarian states have incentives to employ an institutional process to achieve and legitimize their preferred political outcomes. Hence, we expect to see more variation in the number of changes in the size of the legislature after the liberalization of African regimes. For the same theoretical reasons, we exclude non-electoral regimes such as Eritrea, Swaziland, and Somalia, but also states where no data was available (i.e., South Sudan and the Democratic Republic of the Congo).

To estimate the hazard rate of legislature survival (h_i(t)) during time t, we run the following model, where h₀(t) is the baseline hazard function and $β^{'} x$ are the covariates and regression parameters

h_{i} (t) = h_{0} (t) \exp (\begin{array}{l} β^{1} x^{i E x e c u t i v e P o w e r} + β^{2} x^{i E l e c t o r a l V u l n e r a b i l i t y} + β^{3} x^{i I n c r e a s e i n P o p u l a t i o n} \\ + β^{4} x^{i G D P} + β^{5} x^{i Re s o u r c e s} + β^{6} x {{}^{i B i c a m e r a l i s m}+ β}^{7} x^{i P a r t i e s} \\ + β^{8} x^{i P a r l i a m e n t a r y} + β^{9} x^{i D e m o c r a c y} + β^{10} x^{i P R} + β^{11} x^{i D e n s i t y} \\ + β^{12} x {{}^{i E l e c t o r a l C o m m i s s i o n I n d e p e n d e n c e}+ β}^{13} x^{i C o n s t i t u t i o n p r o v i s i o n} \end{array})

o measure legislative expansion, we hand-coded a dummy variable that takes the value of 1 when either the lower and upper houses—or both of them—gain new seats.⁸

The main independent variable is the strength of the executive branch over the legislature and judicial branch, as defined by formal institutions. Operationalizing this power of the executive on other branches is not straightforward, however. First, there are very few cross-sectional measures of executive strength, and when available, these measures often raise concerns. For instance, the Presidential Power Score (Doyle & Elgie, 2016) does not vary across time and codes only 35 African states. Indicators based on expert-survey, such as V-DEM (Coppedge et al., 2021), raise endogeneity concerns as coding experts might use the phenomenon we describe here to inform whether the executive has leverage on the other branches, thus exposing results to reverse causality issues. Second, there is no consensus within the literature regarding the precise constituents of executive strength. If some measures appear to focus on measuring how powerful the executive branch is by looking at the relationship between the executive and other political elites (e.g., Opalo, 2019; Meng, 2020), others have focused on the relative material resources and legitimacy of the executive as proxies for executive power. In fact, a correlation matrix between cross-sectional measures of executive strength (Appendix E) finds that most indicators are not correlated with each other, thus suggesting that they are probably attempting to measure different dimensions of the concept.

With no perfect indicator of executive power at hand, we measure the strength of the executive by looking at whether a presidential term limit was introduced in the constitution. Presidential term limits are often seen as one of the most important formal institutions concerning the executive branch (Posner & Young, 2018). The weakness of constitutional constraints on the executive and the fear of military coups in the post-independence era incentivized African leaders to maximize power and limit the scope of mechanisms of accountability. The third wave of democratization changed this dynamic. Power became more institutionalized, and presidential term limits were introduced in many national constitutions (Reyntjens, 2020; McKie, 2017). Since then, autocrats have used formal channels (i.e., legal loopholes or constitutional amendments) to secure their position. A number of them have managed to formally eliminate the term limits that had previously been introduced, while others have unsuccessfully sought to do so. Using year as our unit of dummy variable accounts for these institutional variations, capturing the executive’s ability to free itself from institutional constraints.

Aside from the fact that the ability to altogether prevent term limits from being established or to eliminate them is symptomatic of executive power, we thus argue that the executive branch is more likely to be stronger in the absence of a presidential term limit for two reasons. First, it offers the opportunity to the president to remain in power endlessly and thus lessens accountability. Second, it is likely that constitutions without presidential term limits lack other strong mechanisms for checks and balances. For instance, in addition to the absence of a presidential term limit, the 1996 Zimbabwean constitution allows the executive branch to appoint Supreme Court Justices and members of the Electoral Commission and call a session of parliament, thus undermining the separation of power (Hatchard et al., 2004). In sum, the presence or absence of presidential term limits offers a good proxy for presidential power and prerogatives.

Because we hypothesize that legislative expansion is less likely to occur under specific institutional configurations, we include these institutional variables. To determine whether the electoral commission operates independently, we use a dummy variable that takes the value of 1 when the executive branch appoints commission members. Because constitutions can have a provision about the parliament size and thus make any attempt to alter the number of seats more difficult, we use a dummy variable to assess the effect of such provisions on legislative expansion.

We introduce some standard controls to test some of the alternative hypotheses discussed above. We control for population growth since it is clearly correlated with legislative size. In addition, country size is usually viewed as a proxy variable for diversity and can test the Taagepera argument that social and cultural diversity are linked to bigger legislatures. GDP per capita and natural resources are included in the mode, both as standard control variables but also because they allow us to test the “patronage” hypothesis since wealthier countries can afford to engage in more aggressive political clientelism. Resnick (2017) has established a link between density and district proliferation in Ghana, so we also control for population density using World Bank Data.

We also control for a set of political variables that could introduce some endogeneity concerns. With scholarship highlighting how the relationship between executives and political elites can shape institutions (Opalo, 2019; Meng, 2020), we hand-coded a dummy variable—electoral vulnerability—that determines whether the executive’s party lost seats in the legislature in that year. We assume that the threat of losing seats in the parliament might alter the relationship with his own backbenchers and incentivize the incumbent to take action to restore a more substantial majority in the house. Because the censoring variable takes the value of 1 for any increase in the lower and upper chamber, we control for bicameralism. Indeed, its recent rise in the region could be a potential confounder for legislative expansion. We include a measure of the effective number of parties to see whether the party system in the country mediates the relationship between the executive and legislative branches, as hypothesized by Jacobs and Otjes (2015). Because presidential systems tend to have a stronger executive branch than parliamentary regimes (Linz, 1990), we control for parliamentarism. Since the executive branch’s strength might be correlated to the level of democracy, we averaged the Freedom House score over the 30 years for each state. Our data suggest that the proclivity toward expansion is lower in states with proportional representation. We hypothesize that political actors have more difficulty benefiting from the expansion under proportional representation and will thus be less motivated to favor expansion. We, therefore, use a dummy variable to control for this confounding factor.

Results

Table 2 presents the results of our Cox Duration Model.⁹ Cox regression coefficients are less easy to interpret than most econometric models. We can, however, use each coefficient’s sign to assess the effect of each variable on the hazard (i.e., risk of changing size). A positive sign indicates that the survival prognosis is worse for states with higher values of that variable. On the other hand, a negative sign suggests that a legislature is less likely to change size. For instance, our results show that legislative expansion is more likely to occur in states with a sturdy executive branch. The same reasoning applies to some of the control variables—electoral vulnerability, number of effective parties, natural resources, electoral commission, and bicameralism. Legislative expansion is more likely to occur when the incumbent’s party loses seats in the legislature. Countries with a bicameral legislature, effective parties, and higher levels of natural resources have a higher risk of seeing their legislature change size. The impact of the dummy variable for bicameralism is particularly noteworthy and suggests that the recent emergence of bicameralism in the region is part of the same dynamic that leads to lower house expansion; Perhaps both houses are the locus of presidential patronage demands; or perhaps, the emergence of bicameralism is related in complicated ways to ongoing executive-legislative relations, and probably to efforts by the president to weaken the legislature.

Table 2.

Survival analysis: Cox Model Coefficients.

	Dependent variable:
	Legislative expansion
Executive power	0.475** (0.212)
Independent Electoral Commission	0.862*** (0.265)
Constitutional Provision	0.168 (0.272)
Electoral Vulnerability	1.562*** (0.234)
Population Growth	0.003 (0.003)
GDP per Capita	−0.00001 (0.00004)
Natural Resources	0.044*** (0.012)
Bicameralism	0.549*** (0.202)
Effective Parties	0.403*** (0.139)
Parliamentary System	0.162 (0.490)
Level of Democracy	−0.042 (0.050)
Proportional Representation	−0.999*** (0.290)
Density	−0.0002 (0.001)
Observation	1,065
Wald Test	103.710*** (df = 13)
LR Test	91.591*** (df = 13)
Score (Logrank) Test	116.116*** (df = 13)

Note: *p < 0.1; **p < 0.05; ***p < 0.01.

Other significant coefficients provide further evidence that legislative expansion is linked to the executive branch’s discretionary power. When presidents are constitutionally responsible for the appointments of electoral commission members, our model indicates that a change in legislature size is more likely to occur. This finding suggests that the level of independence of electoral commissions can foster or undermine the executive’s ability to manipulate the parliament’s size. Proportional representation, on the other hand, is negatively associated with legislative expansion. Because PR systems provides seats in strict proportion to the share of votes, it is harder for the executive to benefit disproportionately from expansion, diluting the benefits of expansion to the incumbent. These results confirm our expectations. To remain in power, African leaders operate within the institutional boundaries of their political system.

A similar argument can be made to explain our counterintuitive results regarding the level of democracy. Our model suggests that legislative expansion is a rarer phenomenon in authoritarian regimes. To better understand these results, we ran the same model using a different subset of the Freedom House score. We excluded democratic states¹⁰ and obtained very similar results to those presented in Table 2.¹¹ Therefore, we argue that electoral autocracies drive the “Level of Democracy” estimate. In other words, we argue that autocrats do not have to rely on strategies such as legislative expansion when their power is both not under constant threat and includes more informal instruments, notably violence. On the other hand, competitive authoritarian leaders need to be more innovative if they want to preserve their rulership while appearing democratic.

Cox regression coefficients are useful to understand the primary trend of the model, but hazard ratios allow us to fully exploit the potential of our Cox Duration Model analysis (see Table 3). Hazard ratios give us the effect size of the covariates. In addition, we can translate the hazard ratio to a probability of a change in the legislature’s size. With the hazard ratio of 1.61, legislatures who operate within a state with a high level of executive discretion almost have a two-time higher chance to see its size change than a legislature in a country where the executive is weaker. When the incumbent loses seats in the parliament, the legislature has a four-time higher chance to see its size being altered. In terms of probability, states with a strong executive have a 61% chance to experience a change of size. In other words, these statistical findings confirm that size variations within African parliaments are not random. The executive branch has engineered the African proclivity toward legislative expansion.

Table 3.

Survival analysis: Hazard ratio.

	Dependent variable:
	Legislative Expansion
Executive Power	1.61
Independent Electoral commission	2.37
Constitutional Provision	1.18
Electoral Vulnerability	4.77
Population Growth	1.00
GDP per Capita	1.00
Natural Resources	1.05
Bicameralism	1.73
Effective Parties	1.50
Parliamentary System	0.18
Level of Democracy	0.96
Proportional Representation	0.37
Density	1.00
Observation	1,065
Wald Test	103.71*** (df = 13)
LR Test	91.59*** (df = 13)
Score (Logrank) Test	116.12*** (df = 13)

Note: *p < 0.1; **p < 0.05; ***p < 0.01.

We also obtain similar and significant results when running a shared frailties model with a gamma distribution fit, which excludes concerns of event times correlation (Appendix A).

Case Studies

Identifying the institutional factors that explain increases in the size of the African legislature does not tell us much about African executives’ specific motivations for doing so. To better understand the executive branch’s motives and the causal mechanisms at play, we conduct three case studies—Uganda, Zambia, and Tanzania. We selected these cases for three reasons. First, the cases needed to allow us to exploit variation on the independent variable—executive strength. Although both Tanzania and Uganda are strong electoral autocracies, the Ugandan executive branch is stronger than the Tanzanian one. Whereas Museveni has been in power for 35 years, Tanzania’s leadership has changed four times since 1990. As we will show, the executive branch is comparatively weaker in Zambia due to a stronger legislature and opposition parties. Second, by including Zambia, we exploit variation on our dependent variable—legislative expansion. Whereas Uganda and Tanzania both underwent significant expansions, the number of seats in Zambia remained mostly constant over the 30-year period. Finally, these three cases allow us to confirm our initial hunch: this process of enlargements cannot be attributed to a single factor. Although each case study shows a similar pattern, they also suggest that each executive branch has different motives, ways to instrumentalize the legislature’s size, and institutional constraints. Using process-tracing, our goal is to bring more granularity to our analysis in order to address the flaws of any cross-sectional observational study.

Tanzania

Tanzania has undergone some political liberalization during the course of the last two decades, albeit under the careful management of the dominant former single party in power since independence, the Chama Cha Mapinduzi (CCM) (Hoffman & Robinson, 2009; Morse, 2014; Collord, 2018). Since the retirement of founding president Julius Nyerere, the party has strictly enforced a two-term limit for the presidency. We argue here that the progressive emergence of an electoral opposition has led the executive to use various institutional mechanisms to protect the CCM’s majority. In addition to the resources of the state and the state-controlled media, from which it clearly benefits, the CCM has also increasingly enjoyed various built-in electoral advantages, such as malapportionment: in 2015, the average CCM legislator represented 119,000 citizens, compared to 152,000 citizens for opposition legislators (Mtega Blog, 2015; Keith Weghorst, personal communication, 2019).

The state has also resorted to legislative expansion to protect its majority. Thus, since the onset of multiparty electoral competition, the National Assembly has been expanded from 284 seats in 1990 to 384 in the 2020 elections. The Tanzanian legislature has long been composed of three different types of seats: (1) elected representatives, (2) gender quota seats, and (3) discretionary appointees filled by the executive. The executive branch has adopted different strategies involving all three of these types of seats in order to weaken the legislature and fulfill its political and electoral agenda.

While a gender quota may well be designed to promote gender equality and improve democratic representation, it can also be a useful tool for an executive branch seeking to weaken the legislature. Pursuant to the Tanzanian constitution, at least 30% of the legislative seats should be reserved for women, indirectly selected by political parties in proportion to their share of electoral votes. However, the evidence suggests that improving women’s representation in the legislature was not the sole effect of this constitutional reform. Because political parties completely control the nomination procedure, elected women often serve as the “agent” of the party (Killian, 1996). Whereas MPs linked to a constituency have an incentive to act for their electorate to be reelected, women MPs have an interest in satisfying the party who appointed them (Killian, 1996). In other words, women appointed by the CCM are more prone to be loyal to the party and secure a more cohesive ruling party. In that regard, these reserved seats provide the significant advantage of increasing ruling party discipline.

Twenty-six new constituencies were created a few months before the 2015 legislative election. To dissipate suspicions of executive interference, the government argued that this decision was taken independently by the National Electoral Commission (NEC) pursuant to a constitutional provision. Although factually true, evidence suggests that the executive branch was behind this decision. First, the NEC hardly operates completely independently since its members are directly appointed by the president, and several elements suggest the existence of a clientelistic relationship between the executive branch and the NEC. The Commission personnel is often abruptly changed with no apparent justification. In fact, the Director of Elections was replaced a few months before the 2015 election. Second, although the government argues that this electoral reform took place within the timeframe set up by the constitution (i.e., every 10 years), the 2015 review took place less than five years after the previous creation of new constituencies. Finally, the 2015 election was the most disputed election since the 1990s. Although the CCM won the majority, the opposition won a larger share of seats than usual. Aware of this danger, the CCM created new seats in its strongholds to prevent the opposition from winning the majority. Hence, one can easily attribute these electoral reforms to the executive branch.

To understand why the executive pushed the NEC to create 26 more constituencies, one needs to look at the location of these constituencies and their political affiliations. On this point, members of the opposition noticed that these new constituencies were located in CMM “strongholds” (Weghorst, personal communication). The NEC suggested that the creation of new constituencies would help reduce existing malapportionment, owing mostly to the larger urban districts and the overrepresentation of rural ones. It was suggested that the creation of these new seats offered the opportunity to remedy this situation. However, in fact, the new constituencies did not have much of an effect on the rural/urban disparities. Moreover, the creation of new constituencies a couple of months before the election created uncertainty among voters about the areas where they should vote and undermined the ability of the opposition to campaign effectively. It is therefore without surprise that the CCM won the majority of these new constituencies. In sum, the Tanzanian case provides strong evidence that the executive branch increased the number of seats in 2015 for political and electoral purposes.

Uganda

Uganda is the least democratic of our cases and has undergone the most autocratization of our three cases (Pike, 2019; Kasfir, 1998; Mugaju and Oloka-Onyango, 2000; Tapscott, 2021; and Tripp, 2010); President Museveni and his National Resistance Movement (NRM) came to power in 1985 following a long insurgency, and initially instated a relatively benign no-party system, arguing that party politics had resulted in the ethnic conflict that led to state collapse in the late 1970s and early 1980s. Under pressure, he agreed to a return to multiparty rule in 2005, but this ostensible act of political liberalization coincided with a progressive hardening of executive authority and Museveni’s own growing intolerance of dissent. Presidential term limits were eliminated in 2005, and the independence of both the judiciary and the executive branches of governments was weakened (Mwenda, 2007, p. 24). The regime has increasingly relied on both patronage and repression to maintain power, though multiparty elections have been convened regularly.

The last parliament before Museveni’s arrival in power resulted from the 1980 general elections and included some 126 seats. The Museveni regime has since pursued a consistent strategy of increasing the size of the legislature, from 278 members in 1989 to 296 members in 2001, 375 seats in 2011, and 426 seats in the current house.¹² Uganda was the first country in Africa to include reserved seats for women in the 1989 elections, and they now total some 112 members in the current legislature. In addition, the 1989 elections witnessed the appointment of some 25 other parliamentarians, including 10 to represent the military and five each to represent the disabled, youth, and labor interests, which have continued to feature in every legislature since then.

Although we do not dispute the representational benefits of these reserved seats, there can be little doubt that the net effect of these different forms of expansion has favored the government and the NRM. Both the opposition and outside experts have argued that the creation of new electoral constituencies has exacerbated both malapportionment and gerrymandering. Even as economic growth has slowed in recent years, popular protests against the regime have increased, the NRM has managed to maintain a comfortable majority in the legislature. In 2011, the NRM’s vote total of 49% yielded 68% of the seats, for example. Indeed, as Africa Confidential put it in February 2016, “Museveni sees a subservient Parliament as vital to maintaining his authority, particularly on key issues such as the oil agenda and governance… an unruly legislature elected in 2011 saw repeated efforts to limit the president’s powers and increase the transparency of public finances. He is determined to prevent a repetition” (February 5, 2016).

The opposition has complained that the creation of new districts has exacerbated the government’s advantage. Redistricting has allowed the government to engage in gerrymandering that has often favored the NRM, with the opposition being relegated to redefined urban areas or areas of opposition support being diluted in multiple constituencies while opposition constituencies are often above average in population.¹³ The European Union’s 2016 election observation mission concluded that “prior to the 2016 elections, 52 new parliamentary constituencies were demarcated. The population ratio was effectively neglected, resulting in profound population discrepancies between electoral constituencies,” pointing out that the average size of a constituency is just under 57 thousand voters, but that constituencies are as small as just over 2000 voters and as large as 200,000 (European Union Election Observation Mission, 2016).

Thus, district creation has helped the NRM maintain control of the House. Official results probably understate NRM control, moreover, because of the large number of independents in Parliament. These are often candidates who lost in MRM primaries but who remain loyal to the movement (Muriaas & Wang, 2012, p. 335n). In addition, it should be noted that the NRM enjoys a higher rate of success among the women reserved seats and particularly the set-aside appointments, which do not include a single opposition party member. The 10 military seats are officially non-partisan, which military members defend by arguing “we side with the government, not the party” (quoted in Muriaas and Wang, p. 330). In sum, on key votes within the House, the NRM majority is handsomely enhanced by the combination of set aside and women members on the one hand and by the process of enlargement of the legislature on the other.

Observers have noted that the regime’s strategy within the legislature has to be understood within the context of its efforts to maintain close NRM control over local politics across the territory through a combination of patronage and hard-edged dominance. According to Wilkins and Vokes (2018), “the fusion of the NRM to the state architecture at the local level—through the local council system and the deployment of resident district commissioners and district intelligence officers—prevents opposition parties from organizing effectively and establishing structures across the country.”

By keeping close control over local politics, legislative expansion has also helped Museveni deal with intra-party dissent. Despite being in power for more than 30 years, some NRM MPs have repeatedly challenged Museveni’s dominance during parliamentary sessions (Collord, 2021). Adding seats to the legislature allows Museveni to reinforce his dominance within the party by increasing the competition between elites. Being an NRM member does not necessarily lead to a seat in the legislature. In 2016, 58% of the NRM lost their parliamentary seats; 88% of them lost against another NRM candidate in the party primary (Wilkins, 2021). Because competition is fierce, party backbenchers thus focus more on their reelection than on challenging Museveni’s dominance.

Thus, legislative expansion in Uganda serves to weaken the electoral opposition and reinforce the presidential majority, but it also strengthens presidential control of the NRM by managing intra-elite dissent and affording new opportunities for patronage to NRM elites in a manner that reinforces the NRM’s power at the local level.

Zambia

Zambia has only experienced very few legislative expansions. Despite numerous attempts to increase the number of seats, the executive branch’s endeavors remained mostly unsuccessful. In the six most recent elections, the number of elected seats increased only twice. These two expansions share a similar pattern to the Tanzanian and Ugandan cases. The most significant increase of seats was done ahead of the 1991 elections while the country was still transitioning from authoritarianism to a multiparty regime. By adding 25 new seats to the Zambian parliament, Kaunda hoped to influence the outcome of the first multiparty election (Boone & Wahman, 2015). The election commission, whose members were appointed by the executive branch, created the new 25 constituencies in rural areas rather than in the already under-represented Lusaka district, thus reinforcing an existing strong rural bias. In fact, the governing UNIP would be routed in the 1991 elections, winning only 25 seats, so Kaunda’s gambit failed to save his regime.

The second increase took place before the 2016 election. Following the recommendations of the electoral commission, six new constituencies were created to “cater for the six new districts whose boundaries cut across constituency boundaries resulting in those constituencies belonging to more than one district.”¹⁴ In its press release, the government underlined that all political parties agreed to the reform. The Electoral Commission Vice Chairperson also stressed how this expansion would increase democratic representation and improve service delivery and voter access.¹⁵ Qualified of “political interference with the order of constituencies in the country,” the opposition vehemently questioned this rationale.¹⁶ Reports from civil society confirmed the opposition’s suspicion by showing how this expansion has exacerbated malapportionment¹⁷ and gerrymandering. For instance, a densely populated district, such as Lusaka, had only two more constituencies than the lowest populated province, North Western.¹⁸

Outside of these two cases, the Zambian legislature’s size remained mainly stable between 1991 and 2016, with several attempts from the executive branch to increase the number of seats failing. For instance, in 2010, the ECZ proposed the creation of 85 new constituencies and an increase to 235 seats.¹⁹ The proposal, however, failed to gain traction in parliament. Hence, in 25 years, only seven appointed and one elected seat were created. Why did the executive branch fail to go through with these reforms? We provide evidence that other formal institutions constrained the government’s action, thus preventing the executive branch from imposing legislative expansion.

Compared to Uganda and Tanzania, Zambia has a recent history of both stronger opposition parties and a weaker government party. Zambian opposition parties have been mostly able to operate effectively since the introduction of the multiparty election system in 1991. Since the return of multiparty politics in 1991, the country has had three main parties—Movement for Multiparty Democracy (MMD), the Patriotic Front (PF), and the United Party for National Development (UPND)—and has witnessed alternations in power. In 2001, the MMD managed to retain the presidency but lost its absolute majority in the Assembly. Indeed, the party of the president last enjoyed a comfortable majority in the legislature in 1996. In 2011, with the second presidential alternation in power, the victory of PF leader Michael Sata to the presidency, the MMD lost much of its support, while the UPND emerged as the leader of the opposition. Finally, the UPND’s candidate, Hakainde Hichilema, defeated Edgar Lungu in the 2021 presidential election for a third presidential alternation since 1991.

Such a fluid environment is not conducive to legislative expansion for two reasons. First, opposition parties can expose and challenge more easily the manipulation of legislative size within the legislature. Second, the effectiveness of opposition parties creates more uncertainty for the incumbent. In a more competitive and fluid political environment, creating new seats could benefit the opposition and backfire against the incumbent. This particular situation occurred in 1991 when the incumbent Kaunda lost the general election against Chiluba and the MMD, despite having increased the legislature.

A second important institutional feature in the Zambian case is the strength of its parliament. Since Kaunda’s defeat, the Zambian parliament has successfully mobilized against legislative expansion. Although Zambia’s Third Republic is marked by a profound imbalance favoring the President (Burnell, 2003), the unicameral chamber resisted several times against the executive’s interferences. In 2011, a coalition led by the UPND successfully blocked a constitutional reform that sought to create 111 new elected seats.²⁰ Three years later, President Sata asked the parliament to create 85 new seats to improve “democratic representation.”²¹ Once again, the parliament opposed the reform on the ground that such an expansion would be costly²² and increase the rural bias.²³ To defeat the proposal, deputies argued that the ECZ used outdated census data to delimit the new constituencies, thus forcing the government to back down.²⁴ In the Third Republic, parliament had asserted its power as early as 2000 when it rejected Chiluba’s bid for a constitutional reform to allow him to run for a third term, with MMD backbenchers overwhelmingly voting against it (Lumina, 2020).

In addition to these two institutional hurdles, the executive branch will need to change the constitution to increase the legislature’s size. Pursuant to Article 68 of the 1991 constitution, the size of the parliament cannot exceed 168 seats. Hence, to amend this provision and add seats to the parliament, the executive branch must introduce a bill “supported by the votes of not less than two-thirds of all the members of the Assembly” in both the second and third readings.²⁵ These constitutional requirements, coupled with a strong parliament and mobilized opposition, make any attempt to manipulate the legislature’s size difficult for the executive branch.

This body of evidence suggests that legislative expansion is less likely to occur when other formal institutions, such as the party system or the legislature, are strong enough to constrain the executive branch’s power. The addition of six newly elected members in 2016 correlates with the deterioration of democratic institutions in the country. Following the 2014 debacle, the PF took action to prevent the legislature from opposing any future expansion. PF Secretary General Wynter Kabimba, who had come out against the 2014 legislative expansion, was removed from office. The PF strategy to incentivize opposition members to switch parties has also undermined the balance of power in the National Assembly. Hence, by weakening the opposition and the legislature’s agency, the PF reinforced its own power, but also the branch’s capacity to impose legislative expansion.

Conclusions

In this paper, we have taken on an issue that has largely been ignored by political scientists: why do some countries regularly increase the number of seats in their legislature? With the help of a new data set examining African legislative expansions going back to 1990, we have shown that in the electoral autocracies and relatively weak democracies of Africa, such increases result from the actions of the executive branch of government and serve largely to weaken the legislature and or provide patronage for the government. Concerns for improving the representation of minorities and women do not fully explain these increases, though they may, in fact, increase their numbers within certain parliaments. Nor do factors specific to the legislature, such as the party system and the leverage of small parties. Instead, it is the ability of the executive branch of government to impose its way in the legislature, thanks to formal institutional powers, that explains legislative expansion.

This paper also contributes to the field of legislative studies and executive-legislative relations in electoral autocracies, which has so far ignored issues of legislative size, despite its evident implications for institutional effectiveness and for the legislature’s ability to undertake its core functions, including oversight of the executive.

Our findings reinforce the view that formal institutions matter in sub-Saharan Africa. Contrary to much conventional wisdom, African political leaders find it useful to manipulate formal institutions for their own advantage and create an impression of legality around their regime. In turn, their ability to do so is conditioned by pre-existing formal rules, some of which appear unusually resilient. Thus, our finding that PR electoral rules significantly lower the chance of a legislature being expanded suggests that Presidents are not free to change electoral rules in order to gain better control of the legislature.

The powerful influence we find of the executive to shape the legislature is a measure of the disproportionate relative power of the former in African electoral autocracies. On the one hand, these regimes have granted enough genuine constitutional prerogatives to the legislature that it can provide a significant check on presidential power. In more autocratic regimes, presidents would not need to stack the chamber with more pliant backbenchers. On the other hand, these regimes do have executives with the political and institutional leverage to influence the internal working of the legislature, unlike in genuinely democratic regimes in which legislative autonomy is more fully protected.

Before the wave of democratization in the early 1990s, we suspect that patronage considerations motivated many of the decisions to increase the size of the legislature, as presidents did not worry about the leverage of the opposition within the legislature and were more attracted by the patronage possibilities afforded in legislative offices, much in the same way they relied on other public offices to maintain political support. Since the return of multi-rule in the early 1990s, we find that patronage opportunities combine with the need to shore up the incumbent party’s support in the legislature, as well as to weaken the opposition and finally to weaken the legislature itself. Our case studies suggest that all four of these motivations co-exist, and we are unable to fully separate them out.

How far does our thesis carry? Our data from other regions of the world suggest that changes in the size of legislatures are a characteristic of young regimes that have recently democratized. It is tempting to argue that the manipulation of legislative size is a characteristic of electoral autocracies all over the world. Nonetheless, our data suggests the predilection for regular increases in Africa is relatively exceptional. Future research should attempt to explain if this is the case and why.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Thalia Gerzso

Notes

Shared Frailty with Gamma Fit Distribution.

	Dependent variable:
	Legislative expansion
Executive power	0.44* (0.23)¹
Independent electoral commission	0.95*** (0.36)
Constitutional provision	0.19 (0.35)
Electoral vulnerability	1.49*** (0.24)
Population growth	0.002 (0.003)
GDP per capita	−0.00001 (0.00004)
Natural resources	0.05*** (0.02)
Bicameralism	0.50 (0.28)
Effective parties	0.39 (0.17)**
Parliamentary system	0.13 (0.61)
Level of democracy	−0.03 (0.07)
Proportional representation	−1.02*** (0.35)
Density	−0.0002 (0.001)
Frailty (Country)	**
Observation	1,065
Variance of rendom effect	0.2
I-likelihood	-682.8
Concordance	0.726 (se=0.726)
likelihood ratio test	126.8***

Note: *p < .1; **p < .05; ***p < .01.

¹Standard error

Schoenfeld Residuals Test.

Variable	χ ²	p-value
Executive Power	0.12	0.728
Independent Electoral Commission	1.81	0.179
Constitutional Provision	0.54	0.46
Electoral Vulnerability	0.08	0.773
Population Growth	4.82	0.028**
GDP per capita	0.85	0.355
Natural resources	3.51	0.061
Bicameralism	3.63	0.057
Effective parties	0.0005	0.981
Parliamentary System	2.35	0.503
Level of Democracy	0.19	0.663
Proportional Representation	2.35	0.125
Density	0.17	0.679
Global	20.8	0.077

Note: *p < .1; **p < .05; ***p < .01.

A p-value inferior to 0.05 suggests a violation of the proportional hazard assumption. Our “Population Growth” variable might have different effects at different times over the 1-year period. To accommodate non-proportional hazards, we build interactions between our “Population Growth” covariate and time into our main Cox regression model (Allison, 2014; Fox & Weisberg, 2018). Because our interaction term is not statistically significant, we are confident in the validity and robustness of our model.

Variable	χ ²	p-value
Executive power	1.32	0.25
Independent electoral commission	7.50	0.0061***
Constitutional provision	2.23	0.13
Electoral vulnerability	0.03	0.86
Population growth	0.59	0.44
GDP per capita	0.05	0.81
Natural resources	0.32	0.57
Bicameralism	8.26	0.004***
Effective parties	0.59	0.44
Parliamentary system	0.23	0.62
Level of democracy	0.23	0.88
Proportional representation	0.44	0.50
Density	0.26	0.61
Population growth * time	0.59	0.44
Global	26.2	0.02**

Survival Analysis for Authoritarian Regimes: Cox Model Coefficients

	Dependent variable:
	Legislative expansion
Executive power	0.443* (0.237)
Independent electoral commission	0.805*** (0.269)
Constitutional provision	0.246 (0.274)
Electoral vulnerability	1.514*** (0.240)
Population growth	0.002 (0.003)
GDP per capita	−0.00001 (0.00004)
Natural resources	0.044*** (0.013)
Bicameralism	0.626*** (0.214)
Effective parties	0.428*** (0.154)
Parliamentary system	0.613 (0.719)
Level of democracy	−0.148** (0.068)
Proportional representation	−0.446 (0.335)
Density	−0.001 (0.001)
Observation	919
Wald test	92.950*** (df = 13)
LR test	85.592*** (df = 13)
Score (Logrank) test	104.630*** (df = 13)

Note: *p < .1; **p < .05; ***p < .01.

Survival Analysis for Authoritarian Regimes: Hazard Ratio

	Dependent variable:
	Legislative Expansion
Executive power	1.60
Independent electoral commission	2.19
Constitutional provision	1.36
Electoral vulnerability	4.63
Population growth	1.00
GDP per capita	1.00
Natural resources	1.05
Bicameralism	1.63
Effective parties	1.56
Parliamentary system	2.03
Level of democracy	0.85
Proportional representation	0.66
Density	1.00
Observation	919
Wald test	88.89*** (df = 13)
LR test	82.24*** (df = 13)
Score (Logrank) test	99.99*** (df = 13)

Note: *p < .1; **p < .05; ***p < .01.

Correlation Between Executive Power Measures.

This correlation matrix includes the main cross-sectional measures of executive power:

Term Limit (TL) is our own measure for executive power. The dummy variable indicates the absence of a term limit for each year (1990–2019)

XCONST (Polity): “refers to the extent of institutionalized constraints on the decision-making powers of chief executives.” This variable covers the 1990–2018 time period.

Executive Power Index (CPI): developed by Elkins et al. (2012), this index “ranges from 0 to 7 and captures the presence or absence of important aspects of lawmaking” (e.g., initiate legislation, decrees, constitutional amendments, declare the state of emergency, challenge the constitutionality of legislation, dissolve the legislature and veto power). This index does not vary across time.

Presidential Power Scores (PP1 et PP2): Developed by Dolye and Elgie (2016): Both scores are calculated across existing datasets and then normalized.

VDEM – Legislative Constraints on the executive (VDEM): This dummy variable measures to what extent the legislature and government agencies, e.g., comptroller general, general prosecutor, or ombudsman capable of questioning, investigating, and exercising oversight over the executive? (1990–2019).

Author Biographies

Thalia Gerzso is a graduate student at the Government Department at Cornell University.

Nicolas van de Walle is Maxwell M. Upson Professor of Government Department at the Government Department at Cornell University.

References

Albaugh

E. A.

(2011). An autocrat’s toolkit: Adaptation and manipulation in ‘democratic’Cameroon. Democratization, 18(2), 388–414. https://doi.org/10.1080/13510347.2011.553361.

Arriola

L. R.

(2009). Patronage and political stability in Africa. Comparative Political Studies, 42(10), 1339–1362. https://doi.org/10.1177/0010414009332126.

Ayee

J. R.

(2013). The political economy of the creation of districts in Ghana. Journal of Asian and African Studies, 48(5), 623–645. https://doi.org/10.1177/0021909612464334.

Bartlett

(2014) Enlarging the house of representatives. New York Times

Benoit

(2004). Models of electoral system change. Electoral studies, 23(3), 363–389. https://doi.org/10.1016/s0261-3794(03)00020-9.

Bleck

Van de Walle

(2018). Electoral politics in Africa since 1990: Continuity in change. Cambridge University Press.

Boix

(1999). Setting the rules of the game: the choice of electoral systems in advanced democracies. American Political Science Review, 93(3), 609–624. https://doi.org/10.2307/2585577.

Boone

Wahman

(2015). Rural bias in African electoral systems: Legacies of unequal representation in African democracies. Electoral Studies, 40(December 2015), 335–346. https://doi.org/10.1016/j.electstud.2015.10.004.

Brownlee

(2007). Authoritarianism in an age of democratization (pp. 124–126). Cambridge University Press.

10.

Burnell

(2003). Legislative-executive relations in Zambia: Parliamentary reform on the agenda. Journal of Contemporary African Studies, 21(1), 47–68. https://doi.org/10.1080/02589000305458.

11.

Colomer

J. M.

(2005). It's parties that choose electoral systems (or, Duverger's laws upside down). Political Studies, 53(1), 1–21. https://doi.org/10.1111/j.1467-9248.2005.00514.x.

12.

Coakley

(2014). The strange revival of bicameralism. The Journal of Legislative Studies, 20(4), 542–572. https://doi.org/10.1080/13572334.2014.926168.

13.

Collord

(2018). The legislature: Institutional strengthening in dominant-party states. Institutions and Democracy in Africa: How the Rules of the game shape political developments (pp. 281–303).

14.

Collord

(2021). Pressuring MPs to act: parliament, organized interests and policymaking in Uganda and Tanzania. Democratization, 28(4), 723–741. https://doi.org/10.1080/13510347.2020.1862088.

15.

Coppedge

Gerring

Carl

H. K.

Lindberg

S. I.

Jan

Alizada

Altman

Bernhard

Cornell

Steven Fish

Gastaldi

Gjerløw

Glynn

Allen

Garry

Ilchenko

Krusell

Anna

Maerz

S. F.

Marquardt

K. L.

Kelly

Mechkova

Medzihorsky

Paxton

Pemstein

fi-Pernes

Johannes von

Brigitte

Sigman

Svend-Erik

Staton

Aksel

Ei-tan

Wang

Tore

Wilson

Ziblatt

(2021). ”V-Dem [Country–Year/Country–Date] dataset v11.1” varieties of democracy project. V-Dem

16.

Doyle

Elgie

(2016). Maximizing the reliability of cross-national measures of presidential power. British Journal of Political Science, 46(4), 731–741. https://doi.org/10.1017/s0007123414000465.

17.

European Union Election Observation Mission (2016). Final Report: Uganda presidential, parliamentary and local council elections 18 february 2016. The European Union.

18.

Gandhi

Przeworski

(2006). Cooperation, cooptation, and rebellion under dictatorships. Economics & Politics, 18(1), 1–26. https://doi.org/10.1111/j.1468-0343.2006.00160.x.

19.

Green

(2010). Patronage, district creation, and reform in Uganda. Studies in Comparative International Development, 45(1), 83–103. https://doi.org/10.1007/s12116-009-9058-8.

20.

Grossman

Lewis

J. I.

(2014). Administrative unit proliferation. American Political Science Review, 108(1), 196–217. https://doi.org/10.1017/s0003055413000567.

21.

Hassan

(2016). A state of change: District creation in Kenya after the beginning of multi-party elections. Political Research Quarterly, 69(3), 510–521. https://doi.org/10.1177/1065912916653476.

22.

Hassan

Sheely

(2017). Executive–legislative relations, party defections, and lower level administrative unit proliferation: Evidence from Kenya. Comparative Political Studies, 50(12), 1595–1631. https://doi.org/10.1177/0010414016679179.

23.

Hatchard

Ndulo

Slinn

(2004). Comparative constitutionalism and good governance in the commonwealth: An eastern and southern african perspective. Cambridge University Press.

24.

Herman

(2015). Parliaments of the world: A reference compendium. Walter de Gruyter GmbH & Co KG.

25.

Hoffman

Robinson

(2009). Tanzania's missing opposition. Journal of Democracy, 20(4), 123–136. https://doi.org/10.1353/jod.0.0117.

26.

Hyden

(2012). African politics in comparative perspective. Cambridge University Press.

27.

Jacobs

Otjes

(2014). Explaining reforms of assembly sizes: Reassessing the cube root law relationship between population and assembly size. Ecpr

28.

Jacobs

Otjes

(2015). Explaining the size of assemblies. A longitudinal analysis of the design and reform of assembly sizes in democracies around the world. Electoral Studies, 40(December 2015), 280–292. https://doi.org/10.1016/j.electstud.2015.10.001.

29.

Jones

B. S.

Branton

R. P.

(2005). Beyond logit and probit: Cox duration models of single, repeating, and competing events for state policy adoption. State Politics & Policy Quarterly, 5(4), 420–443. https://doi.org/10.1177/153244000500500406.

30.

Kasfir

(1998). African ambiguities:"No-party democracy” in Uganda. Journal of Democracy, 9(2), 49–63. https://doi.org/10.1353/jod.1998.0029.

31.

Killian

(1996). A policy of parliamentary ‘special seats’ for women in Tanzania: Its effectiveness. Ufahamu: A Journal of African Studies, 24(2–3). https://doi.org/10.5070/f7242-3016666.

32.

Kimura

(2013). Political change and territoriality in Indonesia: Provincial proliferation. Routledge.

33.

Kreppel

Amie.

(2014). Legislatures. In: Caramani

(Ed.), Comparative Politics (3rd edition, pp. 113–130). Oxford Univeristy Press.

34.

Linz

J. J.

(1990). The perils of presidentialism. Journal of Democracy, 1(1), 51–69. https://doi.org/10.1353/jod.1990.0059.

35.

Lumina

(2020). Frederick Chiluba’s third presidential term bid in Zambia. In Mangala

(Ed.), The politics of challenging presidential term limits in Africa (pp. 205–226). Palgrave Macmillan. https://doi.org/10.1007/978-3-030-40810-7_7.

36.

Mills

(2011). The fundamentals of survival and event history analysis introducing survival analysis and event history analysis (pp. 1–17). SAGE Publications.

37.

McKie

(2017). The politics of institutional choice across Sub-saharan Africa: Presidential term limits. Studies in Comparative International Development, 52(4), 436–456. https://doi.org/10.1007/s12116-017-9252-z.

38.

Meng

(2020). Constraining dictatorship: From personalized rule to institutionalized regimes. Cambridge University Press.

39.

Morse

Y. L.

(2014). Party matters: The institutional origins of competitive hegemony in Tanzania. Democratization, 21(4), 655–677. https://doi.org/10.1080/13510347.2013.764285.

40.

Mtega Blog (2015). One person, one vote? How many voters does each Tanzanian MP represent? http://mtega.com/2015/05/how-many-voters-per-mp/

41.

Mugaju

Oloka-Onyango

(Eds.), (2000). No-Party democracy in Uganda: Myths and realities. Fountain Pub Limited.

42.

Muriaas

R. L.

Wang

(2012). Executive dominance and the politics of quota representation in Uganda. The Journal of Modern African Studies, 50(2), 309–338. https://doi.org/10.1017/s0022278x12000067.

43.

Mwenda

A. M.

(2007). Personalizing power in Uganda. Journal of Democracy, 18(3), 23–37. https://doi.org/10.1353/jod.2007.0048.

44.

Ochieng’Opalo

(2019). Legislative development in Africa: Politics and postcolonial legacies. Cambridge University Press.

45.

Ochieng’Opalo

(2020). Constrained presidential power in Africa? Legislative independence and executive rule making in Kenya, 1963–2013. British Journal of Political Science, 50(4), 1341–1358. https://doi.org/10.1017/s0007123418000492.

46.

Pike

(2019). Combatants: A memoir of the bush war and the press in Uganda. William Pike.

47.

Posner

D. N.

Young

D. J.

(2018). Term limits: Leadership, political competition and the transfer of power. Institutions and democracy in Africa: How the rules of the game shape political developments. Cambridge University Press (pp. 260–278).

48.

Prempeh

H. K.

(2008). Progress and retreat in Africa: Presidents untamed. Journal of Democracy, 19(2), 109–123. https://doi.org/10.1353/jod.2008.0036.

49.

Resnick

(2017). Democracy, decentralization, and district proliferation: The case of Ghana. Political Geography, 59(July 2017), 47–60. https://doi.org/10.1016/j.polgeo.2017.02.011.

50.

Reyntjens

(2020). Respecting and circumventing presidential term limits in sub-saharan Africa: A comparative survey. African Affairs, 119(475), 275–295. https://doi.org/10.1093/afraf/adz029.

51.

Schedler

(2013). The politics of uncertainty: Sustaining and subverting electoral authoritarianism. OUP.

52.

Schuler

Malesky

E. J.

(2014). Authoritarian legislatures. In: Martin

Saalfeld

Strøm

K. W.

. The oxford handbook of legislative studies. Oxford

53.

Shugart

M. S.

Taagepera

(1989). Seats and votes. New Haven

54.

Stigler

G. J.

(1976). The sizes of legislatures. The Journal of Legal Studies, 5(1), 17–34. https://doi.org/10.1086/467542.

55.

Svolik

M. W.

(2012). The politics of authoritarian rule. Cambridge University Press

56.

Taagepera

(1972). The size of national assemblies. Social Science Research, 1(4), 385–401. https://doi.org/10.1016/0049-089x(72)90084-1.

57.

Tapscott

(2021). Arbitrary states: Social control and modern authoritarianism in museveni’s Uganda. Oxford University Press.

58.

Tripp

A. M.

(2010) Museveni’s Uganda: Paradoxes of power in a hybrid regime (Vol. 6). Boulder, CO: Lynne Rienner Publishers.

59.

Tull

D. M.

Simons

(2017). The institutionalisation of power revisited: Presidential term limits in Africa. Africa Spectrum, 52(2), 79–102. https://doi.org/10.1177/000203971705200204.

60.

Van Cranenburgh

(2009). Restraining executive power in Africa: horizontal accountability in Africa’s hybrid regimes. South African Journal of International Affairs, 16(1), 49–68. https://doi.org/10.1080/10220460902986230.

61.

Van de Walle

(2001). African economies and the politics of permanent crisis, 1979–1999. Cambridge University Press.

62.

Wilkins

(2021). Subnational turnover, accountability politics, and electoral authoritarian survival: Evidence from Museveni’s Uganda. Comparative Politics, 54(1), 149–173(25). https://doi.org/10.5129/001041521X16060530242223

63.

Wilkins

Vokes

(Eds.), (2018). Elections in Museveni’s Uganda. Routledge, Comparative Politics, Volume 54, Number 1, October 2021, pp. 149–173(25).

64.

Xinhua (2014). “Niger: l’augmentation du nombre des deputés fait polémique”. A Niamey.