africa forum journal
a Mauduit Study Forums’ Journal
Remy Mauduit, editor
Changing the Rules:
Institutions, Party Systems, and the Frequency of Constitutional Amendments in Africa
Christian B. Jensen is an associate professor of political science at the University of Nevada, Las Vegas, US.
Michelle Kuenzi is an associate professor in the Department of Political Science at the University of Nevada, Las Vegas, US.
Jonathan-Georges Mehanna is a data scientist with Facebook’s Survey and Demographic Science Division, US.
Jensen, Christian Kuenzi, Michelle & Mehanna, Jonathan (2022) Changing the Rules: Institutions, Party Systems, and the Frequency of Constitutional Amendments in Africa, Africa Spectrum, DOI: 10.1177/00020397221082946.
ABSTRACT
What factors influence constitutional stability in the emerging democracies of Sub-Saharan Africa? This is an important question that has never been addressed in a systematic cross-national study of Africa’s emerging democracies. Using the Comparative Constitutions Project dataset and our original dataset for veto players and party systems, we examine the influence of veto players and party system characteristics on the frequency of constitutional amendments in the electoral periods between 1990 and 2020 in forty-one African countries. Our results provide broad support for the veto players theory. More specifically, we find that as the number of actors needed to approve a constitutional amendment increases, the likelihood a constitution will be amended decreases. The strength of party veto players also influences whether a constitution is likely to be amended. Furthermore, our results underscore the importance of party system characteristics in influencing amendment frequency.
Keywords: Africa, Institutions. Party systems, Constitutional amendments.
Identifying the factors contributing to institutional change is central to understanding the politics of transitioning societies. As African societies move away from “Big Man” rule, governing by fiat, towards more democratic forms, the established rules constraining political competition and the legislative process become more important. An important aspect of the question of institutional change is the frequency of constitutional amendments. Despite increasing awareness of and attention to the centrality of institutional politics to political development in Sub-Saharan Africa (hereafter simply Africa), a comprehensive cross-national study of constitutional amendment frequency in the emerging democracies in Africa has been lacking. Addressing this lack of attention to constitutional change in Africa contributes to the growing awareness of the role institutions play in African politics. This is more than simply a geographic extension of the existing literature on amendment frequency. By testing principals of institutional analysis in Africa, we are subjecting them to a crucial test. If we were to find that institutional approaches yielded different results in the African context, this would undermine the claims of advocates of the institutional approach to the generalisability of their framework.
We present a study of the factors that influence constitutional stability in the electoral periods between 1990 and 2020 in forty-one African countries. We find evidence underscoring the importance of veto players theory and party system characteristics, especially concerning electoral volatility and the age of established parties. As Tsebelis[1] notes, several scholars[2] have argued that institutional factors play only a minor role or no role at all in determining the rate of constitutional amendments. Democratic institutions were established relatively recently in many African countries, and this region comprises relatively few consolidated democracies. Therefore, our study constitutes a difficult test of the claim that institutions matter for the rate of constitutional amendments, rendering our findings even more notable. Furthermore, examining constitutional amendment frequency in Africa is important in its own right.[3] Fombad argued that the increased frequency of amendments undermines constitutional legitimacy. Therefore, expanding our empirical understanding of the factors that contribute to or constrain amendment frequency is an inherently important undertaking.
Why Study Constitutions in Africa?
Africa is an excellent setting in which to examine constitutional amendment frequency. Fombad[4] observes that there have been “three generations of constitution-building” in Africa which include “the colonial ‘constitutions,’ the postcolonial or post-independence constitutions, and the post-1990 constitutions.” Most African countries now have new constitutions or have made significant changes to their pre-1990 constitutions.[5]
Despite the prevalence of constitutional change, many echo Prempeh’s[6] lamentation that, “…the modal African presidency has emerged from the recent round of democratic reforms with its extant powers substantially intact.” In Zambia, for example, many constitutional reforms have taken place, but executive power is still robust.[7] Poteete[8]describes how some constitutional amendments intended to strengthen the parliament or rein in the executive in Botswana have failed to have the intended effects. For example, the Independent Electoral Commission was established through constitutional amendments in 1997 but has not functioned as hoped because of its dependence on the executive.[9]
Although opposition leaders may champion decreasing executive power, their perspective tends to change once in power.[10] The case of Senegal’s President Abdoulaye Wade is illustrative. Upon election, President Wade facilitated the adoption of the 2001 constitution that addressed concerns that had been expressed by the opposition and the public. Wade campaigned on abolishing the Senate and reducing the presidential term to five years. He also opposed former President Abdou Diouf enlarging the National Assembly from 120 to 140 seats. The 2001 constitution established the National Assembly as the only parliamentary chamber, changed the presidential term from seven years to five years, and set the number of seats in the National Assembly to 120. In 2008, however, the Senegalese Constitution was amended to change the presidential term from five back to seven years and to enlarge the National Assembly to 150 seats. The Senate was also reestablished; 65 of the 100 Senators were to be appointed by the president.[11] Fatima Diallo’s[12] paper notes that the 2001 Senegalese constitution was amended over fifteen times during Wade’s tenure in office, mostly in ways intended to allow Wade to maintain power and transfer that power to his son, Karim Wade.[13]
The adoption of constitutionally imposed term limits, however, represents an institutional barrier to the indefinite terms of office that long-defined executive power in Africa. Posner and Young[14] argue that the growing importance of formal institutions is reflected in how executives have reacted to term limits. Most presidents accepted the term limits. Those who did not attempt to change the constitution. “This all points to the emergence of a new equilibrium whereby actors…locate the source of political authority in the constitution rather than in a monopoly on the use of violence”.[15] This assertion is part of a broader literature emphasizing the trend toward more democratic constitutions. This pattern has been observed especially in Namibia and South Africa as well as Kenya,[16] where democratic constitutional reform had previously not met with success.[17] The Kenyan Constitution of 2010 strengthened the legislature,[18] increasing its independence.[19]
The rate of constitutional change for democratic development in Africa has ambiguous implications but most scholars agree they are important. Hessebon[20] sees the desire to correct the shortcomings associated with the previous round of constitutional reforms as driving recent constitutional change. In contrast, Fombad[21] contends that “The frequency with which most African post-1990 constitutions are being revised not only erodes their legitimacy but also impacts negatively on their implementation and is the main cause of constitutional fragility.” Fombad[22]observes that most constitutional reforms, even in more democratic countries, are intended to bolster those in power. For example, Fombad [23]notes that amendments to the South African Constitution established controversial floor-crossing rules likely to hurt small opposition parties but leave the African National Congress (ANC) unscathed. Floor-crossing was abolished, however, once significant divisions in the ANC emerged that made floor-crossing more threatening to the ANC.
Some scholars[24] contend that formal institutions have little influence in African countries.[25]If politics is dominated by informal relationships to the point where rules and institutions are not salient, the incentive to rearrange those rules and institutions declines. Similarly, if politicians cannot anticipate their place in the institutional context, they will not know how best to arrange those institutions for their benefit. In either situation, the incentives to amend a constitution lessen. The rate of a constitutional amendment in African countries suggests that formal institutions matter in political life.
Most African countries are now multi-party, electoral regimes;[26] trends over time, however, are not unambiguously moving towards party system institutionalization (PSI).[27] New parties enter the system often while the electoral support for existing parties can rise and fall dramatically.[28]
As countries transition from pure authoritarian regimes towards more competitive forms, constitutions take on greater importance. It is precisely these countries in transition on which our research focuses. Most research on constitutional change focuses on established democracies.[29] Two particularly important features of this literature are the “rigidity” of the Constitution[30] and its length.[31] Rigidity refers to the difficulty of the amendment process itself. Karlsson[32] finds evidence that an additive index of constitutional features making amendments more difficult is associated with fewer constitutional changes. However, Tsebelis and Nardi[33]find that institutional barriers to the amendment are insufficient to explain the pattern of amendments observed in OECD democracies.[34] Indeed, they find that lengthy constitutions are more likely to be amended than shorter ones.[35]
This research agenda has been extended beyond the OECD democracies to include democracies in other regions especially in Latin America but also Asia and some African states as well.[36]However, most of the attention given to African states has focused on the most established democracies on the continent, especially South Africa. We are interested in examining the institutional and party system constraints on constitutional amendments in a wider range of regimes.
Our interest in constitutional amendments originates from our interest in the salience of political institutions in general in transitioning states in Africa.[37] Much of the research on the increasing importance of institutions in emerging democracies has focused on PSI as a dependent variable.[38]
Despite the strong role of presidents in most African regimes,[39] they often depend on the support of coalitions of political groups to remain in power and implement their agendas. Even in authoritarian regimes, some basis of support is required).[40] Furthermore, because of the fluidity of the party systems, when the legislature’s agreement is required to change the constitution, a coalition of two or more parties will likely be required to pass the amendment. The veto player logic leads us to expect that the more parties there are in the legislature, the more likely a multi-party coalition will be required for passage, making passage less likely.
The Importance of Institutions
In Tsebelis’ approach[41] to understanding political institutions, a veto player is any actor whose agreement is required to change the legislative status quo. These can be institutional veto players such as a president or a chamber in a legislature. They can also be partisan veto players such as a political party that, because of its electoral success, is in a pivotal position in the decision-making process. As the number of political conflicts between veto players increases, changing the status quo becomes more difficult.[42] While much has been written about veto players in developed democracies, little attention has been given to the role of veto players in developing and emerging democracies.[43] We contend that veto players play similar roles in African countries with competitive elections, even where democratic consolidation is incomplete, as they play in more established democracies in other parts of the world. We expect that the number of veto players should limit the frequency of constitutional amendments.
Tsebelis’ argues[44] that constitutions are stable so long as they reflect the preferences of the actors whose agreement is required to change them. When the actors, their preferences, or the issues themselves change, a constitutional amendment is possible. The veto player effect, as it applies to constitutional amendments, is also related to the concept of “rigidity” as discussed above.[45] We explore these concepts with two related hypotheses. First, we examine the number of institutional actors whose agreement is required to change the Constitution. This is a direct measure of the number of institutional veto players. For example, in countries with a bicameral legislature, and where both chambers must independently approve the constitutional amendment, we count each chamber as a discrete veto player. We also consider other institutional veto players such as presidents, sub-national units, or popular referenda when the approval of those actors is required for the passage of an amendment. The total number of veto players thus reflects the total number of institutional actors whose approval is required for the passage of the constitutional amendment.
Hypothesis 1: As the number of actors required to change the constitution increases amendments to the constitution become less frequent.
A related issue to the number of actors required to change the status quo is how close the largest party in the system is to being able to make unilateral changes. A single dominant party supported by several smaller parties is a common pattern in Africa’s new democracies.[46] The more control the largest party has over the legislative process, the more control it will wield over any changes to the Constitution. As the largest party’s share of the seats in the legislature increases, the more concentrated we expect political power to be. As legislative dominance increases, we expect it to be easier to amend the Constitution. And this leads us to our next hypothesis:
Hypothesis 2: As the largest party in the system’s seat share approaches, or exceeds, the threshold for unilateral amendment, amendments will become more frequent.
However, one of the difficulties in analyzing the politics of Africa is that the party systems can be so unstable that party affiliations can be less important than ethnic, regional, or other demographic identities. There is considerable research that underscores the importance of ethnic fractionalization in African politics. Nevertheless, even in countries where ethnic politics dominate, ethnic identity can play a similar role to partisan veto players.[47] Ethnic groups’ political influence likely allows them to block changes, much as partisan veto players would in OECD democracies. To evaluate the possibility that ethnic groups act as veto players, we test the following hypothesis:
Hypothesis 3: More ethnic diversity in the system will make amendments less frequent.
Because African politics is often characterized by shifting party loyalties and the distribution of patronage, we can envision a political world in which an unstable party system leads to parties’ shifting institutional preferences. As preferences for institutional arrangements become more fluid, the frequency of efforts to change those institutional arrangements should increase.
Increased volatility can change parties’ positions in the system altering the strategic context in which parties find themselves. Volatility can result from new parties entering the system, older parties departing or electoral support shifting from one party to another. All of this is especially likely in the African context because so many parties are vehicles for prominent individual politicians. Indeed, it is common for such political actors to rename or completely reconfigure their party organizations to suit new needs and priorities. This, in turn, can lead to increased demand for changes to the institutional context in which the parties operate.
Shifting electoral support can change parties’ preferred electoral rules. However long-lived the party is, this can also change that party’s preferences for institutions. Suppose a party that has been in power for some time experiences declining electoral support. They may wish to lock in policies by putting them in the constitution or change electoral rules to reflect their new strategic position. Similar changes in preferences can occur in opposition or smaller supporting parties as well. In any case, changing electoral fortunes can lead to increased demand for institutional change.[48]
Hypothesis 4: As electoral volatility increases, amendments to the Constitution should become more frequent.
While many African political parties are largely vehicles for the personal political careers of particular politicians, in some African countries, parties have stabilized enough to outlast the public careers of their leaders. Longer-lived parties are defined more by their positions in the country than by their current leaders’ needs of the moment. Such parties may pay more attention to institutions. In this context, the age of parties is a reasonable indicator of the focus on institutions as opposed to personal politics. Kuenzi and Lambright[49]found some evidence that the average age of parties is associated with improved democratic characteristics. Furthermore, older parties accumulate expertise and experience over time that enable party leaders to have clearer preferences about their position in the system and how best to optimize the constitution to suit their needs. We expect that as parties age, the politicians in them become more confident in the constitutional characteristics they desire.
Hypothesis 5: As the average age of parties increases, amendments to the constitution will be more frequent.
Data and Methods
Our dataset includes constitutional amendment data for forty-one African countries8 that held multi-party legislative elections between 1990, the year largely seen to mark the start of the democratic era in African politics,[50] and 2020 with a country-electoral period as the unit of analysis.[51] We look first at those countries holding continuous multi-party elections regardless of their level of democracy (forty-one countries). To control for regime type, we use the Polity IV score which ranges from −10 to 10 with higher values being associated with more democracy.[52] We transformed this into a measure ranging from 0 to 20. Controlling for the level of democracy does not fully address the fact that some of the countries that hold elections do not qualify as minimally democratic which means that elections are not competitive. Polity scores below the midpoint suggest that governments control elections to a degree that affects party system dynamics.[53] We are interested in examining the determinants of a constitutional amendment in emerging democracies in which elections are not completely controlled by the government. Indeed, the hypotheses put forward in this paper apply best to regimes that are at least minimally democratic. Therefore, following other studies that use a Polity score close to the midpoint as the democracy cutoff [54] inter alia), we restrict our sample to those twenty-eight countries with a Polity score above 10 (that would be a Polity score of 0 on the original scale). As additional robustness checks, we run the models with a cutoff one point above and one point below that is used for our core model.
Our unit of analysis is the time between elections or the election period. We use this as the unit of analysis because many of our variables of interest (largest party size, electoral volatility, etc.) are constant throughout this period. Using year would therefore count the values of such variables. Using election year would omit many of the amendments that take place between elections. The mean number of observed electoral periods per country is 3.1 for the sample of countries with a polity score above 10. The number of elections (and thus electoral periods) and electoral calendars differ among the African countries included in this study ranging from 1 electoral period to 5 per panel. We characterize our dataset as a longitudinal cross-sectional time series (TSCS) with highly unbalanced panels and a limited number of observations per panel. This allows us to accept the critical assumptions of a TSCS model, such as the assumption of pooling (i.e. stacking observations by units). Specifically, this assumes that all units can accurately be captured through the use of a single regression equation at multiple points in time.[55]
Operationalization
The dependent variable, whether the constitution was amended during the electoral period, is derived mainly from the Comparative Constitutions Project (CCP) dataset which we corrected and significantly expanded to correspond with our sample of countries.[56] The CCP distinguishes between several constitutional events: amendments, new constitutions, and interim periods. Our measure of the dependent variable only considers constitutional amendment events since the process and political climate around the introduction of new constitutions is not the same as the process of amending existing constitutions. Additionally, amendments are much more frequent in Africa than the introduction of new constitutions.
The variable for the total number of institutional veto players was derived from the text analysis of all current and former constitutions[57] of the forty-one countries in our dataset. We focused on the constitutional articles establishing the amendment procedure of the constitutions. We applied Tsebelis’ definition of an institutional veto player as any actor established in the constitution whose agreement is required to change the status quo. For example, if the amendment process requires the approval of both houses of a bicameral legislature and the president, then we determine that there are three institutional veto players.
We also include a variable that captures whether there is more than one way to amend a constitution. This dummy variable takes on a value of one if there is more than one way to amend the Constitution and a value of zero if there is only one way to amend the Constitution. For example, a constitution can be amended through a majority vote of the national assembly with the possibility of a presidential veto or a higher threshold super majority vote by referendum without the possibility of a presidential veto. Those two ways of amending the same constitution differ and so would the number of institutional veto players reflected in the previous variable depending on which amendment process we coded.
For those constitutions with several ways to amend, we isolated the “easiest way of amending” the constitution by examining the articles of the constitutions related to the amendment processes. We use criteria that would make the amendment more difficult[58] or easier.[59] These criteria are all based on the veto players’ logic.
First, we assume that the more actors are required to pass the amendment, the more difficult the procedure of amending the Constitution will be. Second, we assume that the more steps in the procedure (even if the same actor decides on more than one step) are required to pass the amendment, the more difficult the procedure. Some amendment procedures require the same actor to approve an amendment twice. For example, a legislature may need to approve twice with an election taking place between the two approvals. In this situation, we treat the two approvals by the same actor as separate approvals by two different actors. Third, we assume that a joint session of a bicameral legislature is an easier hurdle to clear than separate approvals by the two chambers because a joint session would have a single agenda setter, whereas separate sessions for the two chambers would empower separate agenda setters for each chamber. Finally, we assume that higher vote thresholds for passage at any step, whether in a legislative vote or a referendum, make the procedure more difficult.
A key indicator of veto players’ effects in our model is the ratio between the seat share of the largest party and the percentage of seats required to amend the constitution. This is a measure of the ability of the largest party to amend unilaterally or, barring that, how many allies it would need and how powerful would they be relative to the largest party itself. For example, suppose some party has a seat share of 55 percent and the threshold for amending the constitution requires a 60 percent vote. The largest party ratio value in this example is 0.55/0.60 = 0.9167. This indicates that the largest party has roughly 92 percent of the seats required to amend the Constitution. Suppose that the party had 62 percent of the seats under the same threshold. In that example, the party’s ratio score would be 0.62/0.60 = 1.0333. This indicates that they have more than 103 percent of the seats required to amend the Constitution. Finally, consider a party that only has a 40 percent seat share in a system requiring a 60 percent vote to amend. In that situation, the party’s ratio score would be 0.40/0.60 = 0.6667, indicating that the party has roughly 67 percent of the seats required to amend. Under our logic, the constitution would be most likely to be amended in the second example, because the largest party can unilaterally amend. It would be somewhat likely to be amended in the first example because the largest party would need fewer and weaker allies to get over the threshold. The prospects for amendment would be weakest in the third example. Data on the largest party seat share were derived from[60] and then adapted and extended using the Inter-Parliamentary Union’s Parline database, the International Institute for Democracy and Electoral Assistance (IDEA), and the African Elections Database to match the spatial and temporal domains of this study.
To capture ethnic diversity, we employ an index of ethnic fractionalization.[61] There are several measures of ethnic fractionalization for African countries in the literature.[62] The measure developed by Alesina et al. offers the least amount of missing data of the measures reviewed by Posner.[63]
To capture electoral volatility, we employ “Pedersen’s index of electoral volatility, which measures the net change in the seat (or vote) shares of all parties from one election to the next”.[64] This is the standard measure of overall electoral volatility in the literature. Like Ferree[65], Kuenzi et al.,[66] and Weghorst and Bernhard,[67] we use seat shares to calculate our measure of electoral volatility because vote shares are not available for many African elections. We employ the average age of parties as an indicator of party system stability.[68]14 As shown in Table A1 in the online appendix, there is considerable diversity among party systems on this measure with party age values ranging from 4.5 to 53 years. We adapted and extended Kuenzi and Lambright’s[69] data on electoral volatility using the Inter-Parliamentary Union’s Parline database,[70] the IDEA,[71] and the African Elections Database.[72]
We control for the effects of variables that have been linked to a constitutional amendment in the literature. We control for two characteristics of constitutions, the length of the constitution, and the age of the constitution. We also account for the level of democracy in a country. We use a measure of economic growth to control the effects of economic performance. Finally, we control for the length of the electoral period.
Estimation Methods
Since our dependent variable, whether the constitution was amended during an electoral period, is a binary variable and our data are cross-sectional time series with unbalanced panels, we use time series logit with random effects as our estimation technique. Because we are interested in how the variation in factors across countries (units) influences the likelihood that a constitution will be amended during an electoral period, fixed effects are not an appropriate estimation technique for this study. Several of our independent variables of interest, such as the number of institutional veto players, whether there are multiple ways to amend the constitution, and ethnic diversity are time-invariant for the period covered in this study. Moreover, the outcome variable does not vary across electoral periods in some countries. Our diagnostics of the distribution of the error terms indicate that our models have a low level of heteroscedasticity.[73] We address this issue by running our models with robust standard errors.[74]
Results
As can be seen in Table 1, the results of the analysis provide support for many of the hypotheses. In the first column (Model 1), we see the results for all the countries in our sample, including those countries that are relatively authoritarian. In Model 2, the analysis is restricted to those countries with competitive elections.[75] The veto players theory is supported by the results displayed in Table 1. The coefficient for the number of institutional veto players is negative and highly significant in both models indicating that the likelihood a constitution will be amended declines as the number of institutional veto players increases, as predicted in Hypothesis 1. In African countries, as elsewhere, the likelihood of a constitution will be amended declines as the number of institutional actors required to approve the change increases. The post-Cold War period has been one characterized by a rapid spread of democracy. Institutions have been growing in importance across the continent and therefore the institutional arrangements guiding the amendment process have the effects in the African context one would expect based on theory. The veto players theory is also supported by the results for the largest party’s ratio of the threshold for unilateral amendment which is significant and positive in both models. These results are consistent with our second hypothesis.
Table 1. Determinants of Constitutional Amendment.
| Model 1 | Model 2 | |
| Veto players | −0.892** (0.345) | −1.085*** (0.285) |
| Largest party ratio | 1.971* (0.891) | 5.730** (1.937) |
| Ethnic diversity | −1.294 (1.435) | 0.712 (1.434) |
| Electoral volatility | 0.0255+ (0.0142) | 0.0617* (0.0263) |
| Party age | 0.0702* (0.0309) | 0.0824* (0.0399) |
| Executive turnover | −0.801 (0.670) | −1.027 (1.035) |
| Length of constitution | 0.0000493** (0.0000183) | 0.0000695** (0.0000257) |
| Age of the constitution | 0.00580 (0.0261) | −0.0209 (0.0306) |
| Multiple ways to amend | 0.883 (0.589) | −0.184 (0.624) |
| Polity | −0.118+ (0.0715) | −0.0988 (0.137) |
| GDP PC change (t−1) | −0.0352 (0.0324) | −0.0736+ (0.0452) |
| Length of EP (months) | 0.0513* (0.0237) | 0.0458 (0.0342) |
| Constant | −4.322 (2.730) | −9.916+ (5.233) |
| lnsig2u | −0.267 (1.159) | −14.43 (.) |
| Observations | 127 | 77 |
Note: Standard errors are in parentheses.
Significance levels: +p ≤ .10, *p ≤ .05, **p ≤ .01, ***p ≤ .001 for a two-tailed test
The coefficient for ethnic diversity is negative as expected in Model 1 but positive in Model 2. It is not significant in either model, so we cannot conclude that it influences amendment frequency. While hypothesis 3 is not supported, the effects of a country’s ethnic demography may be reflected in other institutions which do influence amendment frequency.
Electoral volatility is associated with a higher probability that a constitution will be amended in both Models 1 and 2, although the coefficient is only significant at the 0.1 level in Model 1 which includes electoral regimes that are more authoritarian. We expected that it would be difficult to see the effects of party system characteristics among regimes in which the governing party can limit competition. The coefficient for electoral volatility is positive and significant in Model 2, in which the analysis is restricted to only those cases where the government does not control electoral competition. It is not surprising that when the sample is restricted to only those cases in which elections are competitive, the effects of electoral volatility become clear. These results support Hypothesis 4. In addition, we find that, as political parties endure, the salience of constitutions increases leading political actors, such as parties, to try to further their agendas via constitutional amendments. The coefficient for the average age of parties is positive and significant in the models. This result supports Hypothesis 5.
Turning to the control variables, we can see that the length of the constitution influences the amendment frequency in Africa’s multi-party systems. Overall, our finding is consistent with that of Tsebelis and Nardi[76]who find that longer constitutions are more likely to be amended in OECD democracies. The coefficient for our measure of democracy, Polity IV, is negative and significant at the 0.1 level in Model 1 and negative but not significant in Model 2. The analysis in Model 2 is restricted to countries in which elections are competitive and thus there is less variation in the level of democracy in Model 2. When the full range of regimes is included in the analysis (Model 1), it appears that constitutions may be less likely to be amended in more democratic regimes. This result is consistent with the idea that the strength of democracy and institutional stability reinforce each other. The coefficient for economic growth is negative in both models and significant at the 0.1 level in Model 2 implying that the system may be less likely to be amended when the economy is performing well. The coefficient for the length of the electoral period is positive in both models but only significant at the 0.1 level in Model 1. Logically, a longer electoral period would provide a greater opportunity for constitutions to be amended. The other control variables appear unrelated to the likelihood of constitutional amendment.
We explore the substantive impact of the covariates of interest on the probability a constitution will be amended by calculating how a change in the value of our covariate of interest affects the probability of a constitutional amendment when the values of the other independent and control variables are set at their means (marginal effects). As can be seen in the plots in Figure 1, the independent variables of interest have significant substantive effects on the likelihood of constitutional amendment. The large substantive effect of the number of institutional veto players on the likelihood a constitution will be amended is clear. Malawi is an example of a country with only one veto player; a super majority of 66 percent of Malawi’s legislative house must approve the amendment. It is therefore not surprising that Malawi’s constitution has been amended many times, including three times during the first three years of the constitution’s existence.
Figure 1. Marginal Effect of Key Predictors of Constitutional Amendments with 95 percent Confidence Levels.
A few examples illustrate the importance of the ratio of the largest party seat share to amending the constitution. Nearly all the countries included in our study have or had presidential term limits. To succeed in amending the constitution to either extend or abolish term limits among the countries in our sample, it appears that having a ratio of 1 or above was of great importance. For example, although many constitutional amendments have been made to Malawi’s constitution, the attempts to extend the presidential term limits during President Muluzi’s last term in office failed when the ruling party did not control the requisite proportion of the seats needed to amend the constitution. Of course, having a ratio of 1 or above cannot guarantee the passage of such a controversial amendment as cohesion in the presidential party would also appear to play a role, which President Chiluba found out in his attempts to amend the Zambian Constitution so that he could run for a third term. Although President Chiluba’s Movement for Multiparty Democracy (MMD) had more than enough seats to pass a constitutional amendment in the legislature, fifty-nine MMD legislators committed to writing to prevent the passage of an amendment of this type.[77] In contrast, President Nujoma had well above the number of seats in the legislature needed to change the constitution and the level of cohesion in his party, SWAPO, to be able to successfully amend the constitution in 1999 so that he could serve an additional term.
The large substantive effect of electoral volatility on the likelihood of constitutional amendment is also evident in Figure 1. Senegal is an example of a country that saw its constitution amended many times after an election that produced a very high level of volatility.
The important impact of party age on the likelihood of constitutional amendment is quite intuitive. Leaders of parties that have endured are likely to have clear ideas of the sorts of constitutional arrangements that are most beneficial to their parties. They are also likely to have mastered the rules of the game and therefore know what strategies are likely to be effective in getting an amendment passed. South Africa has the highest party age in our sample, and not surprisingly its constitution has been frequently amended.
Conclusion
Our results highlight the importance of institutions in constraining political outcomes in African countries. Veto players theory finds strong support in our results. As in other regions, as the number of institutional actors who must agree to a change in the constitution increases, the likelihood that the constitution will be amended declines. We also find that partisan veto players have a powerful effect on the likelihood of constitutional amendment. Taken together, our results indicate that the characteristics of the political party system influence constitutional amendment frequency markedly. As parties become more institutionalized, as reflected in their “age” or years in existence, the constitution’s rules become more and more important. Most of Africa’s democracies are young and therefore the average ages of parties in African countries are lower than those in other parts of the world, even those comprising many emerging democracies such as Latin America.[78] Perhaps when parties reach the age of the parties we see now in long-standing democracies such as those in Western Europe, we will see a change in this relationship. High levels of legislative volatility also encourage constitutional amendment for very different reasons. In systems with high levels of electoral volatility, those in power anticipate losing power and wish to have their preferences enshrined in the constitution for the future. These findings point to the important role political institutions play in influencing political outcomes in Africa. Given the youth of democratic institutions and the small number of consolidated democracies in Africa, this finding is even more notable. Our findings do not mean that informal institutions and relationships are not important. As Cheeseman observes,[79] formal institutions and informal institutions need not be rivals but rather they may have a symbiotic relationship. Some of the findings of our study echo those of studies done on different world regions which supports the notion that certain universal causal processes may be at play when it comes to constitutional change. For example, the rigidity of the constitution, a concept integrally related to veto players’ logic, and the length of the constitution play important roles in determining the rate of a constitutional amendment in Latin America.[80] Although we did not hypothesize about the effect of the length of the constitution, our results also indicate that longer constitutions are more likely to be amended in African countries. Our results point to the importance of the seat share of the largest party in influencing constitutional amendment in the African context. We also find, however, that other party system characteristics play a role.
To the best of our knowledge, this is the first cross-national empirical study to examine the factors that influence constitutional amendment in Africa. Since constitutions are considered the foundation of the rule of law and democracy, understanding the forces that shape them is critical. Our findings indicate a close link between veto players, political party systems, and constitutional amendment. Future research could examine other types of constitutional change in Africa such as the adoption of new constitutions and other factors that space and time did not allow us to explore. Exploring whether there is some type of tradeoff between different types of constitutional change could also be fruitful.
[1] Tsebelis G (1995) Decision making in political systems: Veto players in presidentialism, parliamentarism, multicameralism and multipartyism. British Journal of Political Science 25(3): 289–325. Crossref Google Scholar
[2] Ginsburg T, Melton J (2015) Does the constitutional amendment rule matter at all? Amendment cultures and the challenges of measuring amendment difficulty. International Journal of Constitutional Law 13(3): 686–713. Crossref Google Scholar
[3] We focus exclusively on constitutional amendments. The emergence of new constitutions is an important subject of study but it is a separate one. We are interested in whether amendment frequency is constrained by existing institutional features. Conducting an empirical analysis of the creation of new constitutions would require information about the actors, their preferences, and the rules constraining them that we do not have access to.
Fombad CM (2014) Constitution-building in Africa: The never-ending story of the making, unmaking and remaking of constitutions. African and Asian Studies 13(4): 429–451. Crossref Google Scholar
[4] Fombad CM (2014) Constitution-building in Africa: The never-ending story of the making, unmaking and remaking of constitutions. African and Asian Studies 13(4): 429–451. Crossref Google Scholar
[5] Id.
[6] Prempeh HK (2008) Progress and retreat in Africa: Presidents untamed. Journal of Democracy 19(2): 109–123. Crossref Google Scholar
[7] Hinfelaar M, Wahman M (2021) Electoral turnovers and the disappointment of enduring presidential power: Constitution making in Zambia. Journal of Eastern African Studies 15(1): 63–84. Crossref Google Scholar
[8] Poteete A (2010) Renegotiation of executive powers and executive-legislative relations in Botswana. Unpublished paper presented at the American Political Science Association, 2–5 September 2010, Washington, DC. Google Scholar
[9] Id.
[10] Prempeh HK (2008) Progress and retreat in Africa: Presidents untamed. Journal of Democracy 19(2): 109–123. Crossref Google Scholar
[11] Kuenzi M, Lambright G (2001) Party system institutionalization in 30 African countries. Party Politics 7(4): 437–468. Crossref Google Scholar
[12] Diallo F (2013) Constitutional Developments in Africa: Senegal. Paper presented at a Seminar of the Oxford Constitutions of the World African Reporters. Google Scholar
[13] Fombad CM (2014) Constitution-building in Africa: The never-ending story of the making, unmaking and remaking of constitutions. African and Asian Studies 13(4): 429–451. Crossref Google Scholar
[14] Posner DN, Young DJ (2018) Term limits. In: Cheeseman N (ed) Institutions and Democracy in Africa: How the Rules of the Game Shape Political Developments. Cambridge: Cambridge University Press, pp. 260–277. Crossref Google Scholar
[15] Id.
[16] Ndulo M (2018) Constitutions: The politics of constitutional reform. In: Cheeseman N (ed) Institutions and Democracy in Africa: How the Rules of the Game Shape Political Developments. Cambridge: Cambridge University Press, pp. 117–136. Crossref Google Scholar
[17] Cottrell J, Ghai Y (2007) Constitution-making and democratization in Kenya (2000–2005). Democratisation 14(1): 1–25. Crossref Google Scholar
Ndegwa SN (1998) The incomplete transition: The constitutional and electoral context in Kenya. Africa Today 45(2): 193–211. Google Scholar
[18] Kadima D, Owuor F (2014) Kenya’s decade of experiments with political party alliances and coalitions-motivations, impact and prospects. Journal of African Elections 13(1): 150–180. Crossref Google Scholar
[19] Opalo K (2019) Legislative Development in Africa: Politics and Postcolonial Legacies. New York: Cambridge University Press. Crossref Google Scholar
[20] Hessebon GT (2014) The fourth constitution-making wave of Africa: Constitutions 4.0. Temple International & Comparative Law Journal 28(2): 185–214. Google Scholar
[21] Fombad CM (2014) Constitution-building in Africa: The never-ending story of the making, unmaking and remaking of constitutions. African and Asian Studies 13(4): 429–451. Crossref Google Scholar
[22] Id.
[23] Id.
[24] Bayart J (1993) State in Africa: The Politics of the Belly. London: Longman Publishing Group. Google Scholar
Chabal P, Daloz J (1999) Africa Works: Disorder as Political Instrument. Oxford: James Currey. Google Scholar
[25] Cheeseman N (2018) Introduction: Understanding African politics: Bringing the state back in. In: Cheeseman N (ed) Institutions and Democracy in Africa: How the Rules of the Game Shape Political Developments. Cambridge: Cambridge University Press, pp. 1–38. Crossref Google Scholar
[26] Gyimah-Boadi E (2015) Africa’s waning democratic commitment. Journal of Democracy 26(1): 101–113. Crossref Google Scholar
[27] Kuenzi M, Tuman JP, Rissmann MP, et al. (2019) The economic determinants of electoral volatility in Africa. Party Politics 49(2): 621–631. Crossref Google Scholar
Nevertheless, it is important to remember that the institutionalization of these party systems is not the same thing as democratization (see Hicken and Kahonta, 2011).
[28] Kuenzi M, Tuman JP, Rissmann MP, et al. (2019) The economic determinants of electoral volatility in Africa. Party Politics 49(2): 621–631. Crossref Google Scholar
Kuenzi M, Lambright G (2001) Party system institutionalization in 30 African countries. Party Politics 7(4): 437–468. Crossref Google Scholar
Kuenzi M, Lambright G (2005) Party systems and democratic consolidation in Africa’s electoral regimes. Party Politics 11(4): 423–446. Crossref Google Scholar
Weghorst KR, Bernhard M (2014) From formlessness to structure? The institutionalization of competitive party systems in Africa. Comparative Political Studies 47(12): 1707–1737. Crossref Google Scholar
[29] Karlsson C (2016) Explaining constitutional change: Making sense of cross-national variation among European Union member states. Journal of European Public Policy 23(2): 255–275. Crossref Google Scholar
Lorenz A (2005) How to measure constitutional rigidity: Four concepts and two alternatives. Journal of Theoretical Politics 17(3): 339–361. Crossref Google Scholar
Tsebelis G, Nardi DJ (2016) A long constitution is a (positively) bad constitution: Evidence from OECD countries. British Journal of Political Science 46(2): 457–478. Crossref Google Scholar
[30] Karlsson C (2016) Explaining constitutional change: Making sense of cross-national variation among European Union member states. Journal of European Public Policy 23(2): 255–275. Crossref Google Scholar
Lorenz A (2005) How to measure constitutional rigidity: Four concepts and two alternatives. Journal of Theoretical Politics 17(3): 339–361. Crossref Google Scholar
[31] Tsebelis G (2017) The time inconsistency of long constitutions: Evidence from the world. European Journal of Political Research 56(4): 820–845. Crossref Google Scholar
[32] Karlsson C (2016) Explaining constitutional change: Making sense of cross-national variation among European Union member states. Journal of European Public Policy 23(2): 255–275. Crossref Google Scholar
[33] Tsebelis G, Nardi DJ (2016) A long constitution is a (positively) bad constitution: Evidence from OECD countries. British Journal of Political Science 46(2): 457–478. Crossref Google Scholar
[34] Id.
[35] We control for both concepts in our model and explain how we operationalize them below.
[36] Albert R (2015) Amending constitutional amendment rules. International Journal of Constitutional Law 13(3): 655–685. Crossref Google Scholar
Linder W, Bächtiger A (2005) What drives democratization in Asia and Africa? European Journal of Political Research 44(6): 861–880.Crossref Google Scholar
Negretto GL (2012) Replacing and amending constitutions: The logic of constitutional change in Latin America. Law & Society Review 46(4): 749–779. Crossref Google Scholar
Tsebelis G (2017) The time inconsistency of long constitutions: Evidence from the world. European Journal of Political Research 56(4): 820–845. Crossref Google Scholar
[37] Brambor T, Clark WR, Golder M (2007) Are African party systems different? Electoral Studies 26(2): 315–323. Crossref Google Scholar
Patterson DP, Fadiga-Stewart L (2009) The strategy of party dominant politics: Electoral institutions and elections outcomes in Africa. CEU Political Science Journal 4(2): 252–257. Google Scholar
Posner DN, Young DJ (2007) The institutionalization of political power in Africa. Journal of democracy 18(3): 126–140. Crossref Google Scholar
Reyntjens F (2016) The struggle over term limits in Africa: A new look at the evidence. Journal of Democracy 27(3): 61–68. Crossref Google Scholar
[38] Dalton RJ, Weldon S (2007) Partisanship and party system institutionalization. Party Politics 13(2): 179–196. Crossref Google Scholar
Kuenzi M, Tuman JP, Rissmann MP, et al. (2019) The economic determinants of electoral volatility in Africa. Party Politics 49(2): 621–631. Crossref Google Scholar
Manning C (2005) Assessing African party systems after the third wave. Party Politics 11(6): 707–727. Crossref Google Scholar
Weghorst KR, Bernhard M (2014) From formlessness to structure? The institutionalization of competitive party systems in Africa. Comparative Political Studies 47(12): 1707–1737. Crossref Google Scholar
[39] Several countries in our analysis are semi-presidential systems in which the President is the dominant figure.
[40] Geddes B, Zaller J (1989) Sources of popular support for authoritarian regimes. American Journal of Political Science 33(2): 319–347. Crossref Google Scholar
Geddes B, Wright J, Frantz E (2014) Autocratic breakdown and regime transitions: A new data set. Perspectives on Politics 12(2): 313–331. Crossref Google Scholar
[41] Tsebelis G (1995) Decision making in political systems: Veto players in presidentialism, parliamentarism, multicameralism and multipartyism. British Journal of Political Science 25(3): 289–325. Crossref Google Scholar
Tsebelis G (2002) Veto Players: How Political Institutions Work. New York: Russel Sage Foundation. Crossref Google Scholar
[42] Id.
Note that this is largely consistent with historical institutional approaches such as “path dependency” (e.g. Pierson, 2000). Path dependency argues that past institutional choices constrain future options for change. The veto players approach does not reject that overall view but rather offers a specific mechanism by which past institutional choices make future change more or less likely.
[43] This is partly due to the difficulty in identifying the ideological positions of African political parties; a key measure of the veto player effect. We use a variety of indicators to measure veto player effects in the countries in our sample. We explain these in more detail below.
[44] Tsebelis G (2017) The time inconsistency of long constitutions: Evidence from the world. European Journal of Political Research 56(4): 820–845. Crossref Google Scholar
[45] Karlsson C (2016) Explaining constitutional change: Making sense of cross-national variation among European Union member states. Journal of European Public Policy 23(2): 255–275. Crossref Google Scholar
Lorenz A (2005) How to measure constitutional rigidity: Four concepts and two alternatives. Journal of Theoretical Politics 17(3): 339–361. Crossref Google Scholar
Tsebelis G (2002) Veto Players: How Political Institutions Work. New York: Russel Sage Foundation. Crossref Google Scholar
[46] Bogaards M (2008) Dominant party systems and electoral volatility in Africa: A comment on Mozaffar and Scarritt. Party Politics 14(1): 113–130.Crossref Google Scholar
van de Walle N (2003) Presidentialism and clientelism in Africa’s emerging party systems. The Journal of Modern African Studies 41(2): 297–321. Crossref Google Scholar
[47] Buttorff G (2015) Coordination failure and the politics of tribes: Jordanian elections under SNTV. Electoral Studies 40: 45–55. Crossref Google Scholar
Kuenzi M, Lambright G (2005) Party systems and democratic consolidation in Africa’s electoral regimes. Party Politics 11(4): 423–446. Crossref Google Scholar
McCauley JF (2014) The political mobilization of ethnic and religious identities in Africa. American Political Science Review 108(4): 801–816. Crossref Google Scholar
Posner DN (2004a) Measuring ethnic fractionalization in Africa. American Journal of Political Science 48(4): 849–863. Crossref Google Scholar
Posner DN (2004b) The political salience of cultural difference: Why Chewas and Tumbukas are allies in Zambia and adversaries in Malawi. American Political Science Review 98(4): 529–545. Crossref Google Scholar
[48] All these sources of volatility drive amendment frequency in the same direction. As such examining them in combination is the correct approach. Furthermore, there is precedent for aggregating both types of volatility into a single measure (e.g. Kuenzi and Lambright, 2005; Mainwaring and Zoco, 2007; Pederson, 1983; Tavits, 2005). Furthermore, for methodological reasons that we discuss in more detail below, dividing volatility into two distinct measures is problematic in the African context.
[49] Kuenzi M, Lambright G (2005) Party systems and democratic consolidation in Africa’s electoral regimes. Party Politics 11(4): 423–446. Crossref Google Scholar
[50] Bleck J, van de Walle N (2019) Electoral Politics in Africa Since 1990: Continuity and Change. New York: Cambridge University Press. Google Scholar
[51] An electoral period is a time between two consecutive elections. Electoral periods disrupted by democratic reversals because of coups or civil wars are therefore excluded from the analysis
[52] Marshall MG, Gurr TR, Jaggers K (2017) PolityIV Project: Political Regime Characteristics and Transitions, 1800-2016. Vienna: Center for Systemic Peace. Google Scholar
[53] Mainwaring S, Zoco E (2007) Political sequences and the stabilization of interparty competition: Electoral volatility in old and new democracies. Party Politics 13(2): 155–178. Crossref Google Scholar
[54] Butkiewicz JL, Yanikkaya H (2007) Time-consistent polities and growth in developing countries: An empirical analysis. Review of World Economics 143(2): 306–323. Crossref Google Scholar
Mainwaring S, Zoco E (2007) Political sequences and the stabilization of interparty competition: Electoral volatility in old and new democracies. Party Politics 13(2): 155–178. Crossref Google Scholar
[55] Mainwaring S, Zoco E (2007) Political sequences and the stabilization of interparty competition: Electoral volatility in old and new democracies. Party Politics 13(2): 155–178. Crossref Google Scholar
[56] We rely on several sources to identify constitutional amendments. We focus on official government online records, national assembly records, and other legal and archived documents that we translated from French, Arabic, Portuguese, Swahili, and other local languages. Where official government records could not be found, we relied on other non-legal but reliable sources such as press releases, official newspaper records, and amendment voting records. We identified and corrected both omissions (missing amendments) as well as errors (incorrect amendments identification) in the CCP dataset. We then updated the CCP dataset to match the temporal domain of our study (1990 to 2010) as the CCP did not cover all the years that we needed. Finally, we expanded the CCP dataset to include all the countries in our sample.
[57] We employ Optical Character Recognition (OCR) tools to conduct an in-text analysis of the articles establishing the amendment of the constitutions. These include both the introductory process of amending articles, the conditions of amending as well as any final signatories to the amendments such as presidential veto powers or the override of that veto thereafter.
[58] Examples include higher thresholds to win a vote either in referenda or in the legislatures, more actors required to approve the amendments, and more steps required to approve the amendments even if some steps are repeated approvals from the same actors.
[59] Examples include lower thresholds to win a vote, fewer different actors required to approve the amendments, and a joint session of a bicameral legislature instead of separate approvals by each chamber of the bicameral legislature.
[60] Kuenzi M, Tuman JP, Rissmann MP, et al. (2019) The economic determinants of electoral volatility in Africa. Party Politics 49(2): 621–631. Crossref Google Scholar
[61] Alesina A, Devleeschauwer A, Easterly W, et al. (2003) Fractionalization. Journal of Economic Growth 8(2): 155–194. Crossref Google Scholar
[62] Posner DN (2004a) Measuring ethnic fractionalization in Africa. American Journal of Political Science 48(4): 849–863. Crossref Google Scholar
[63] Id.
[64] Mainwaring S, Scully T (Eds) (1995) Building Democratic Institutions: Party Systems in Latin America. Stanford: Stanford University Press. Google Scholar
[65] Ferree KE (2010) The social origins of electoral volatility in Africa. British Journal of Political Science 40(4): 759–779. Crossref Google Scholar
[66] Kuenzi M, Tuman JP, Rissmann MP, et al. (2019) The economic determinants of electoral volatility in Africa. Party Politics 49(2): 621–631. Crossref Google Scholar
[67] Weghorst KR, Bernhard M (2014) From formlessness to structure? The institutionalization of competitive party systems in Africa. Comparative Political Studies 47(12): 1707–1737. Crossref Google Scholar
[68] Cruz C, Keefer P, Scartascini C (2016) The database of political institutions 2015. Crossref Google Scholar
[69] Kuenzi M, Lambright G (2001) Party system institutionalization in 30 African countries. Party Politics 7(4): 437–468. Crossref Google Scholar
[72] www.africanelections.tripod.com
[73] See Figure A1 in the online appendix for the plot of the normality of residuals. Additional tables in the appendix examine alternative measures of ethnic diversity and democracy as well as different cut off points for the level of democracy of countries included in the analysis. Finally, the appendix includes country dummy variables for three key cases. In all the alternative models our substantive results largely held.
[74] Multicollinearity, however, is not a problem in our analysis. The highest variance inflation factor (VIF) for any of the variables included in the model is 1.87 for the level of democracy while the overall mean VIF for the model is 1.41.
[75] As noted, our cutoff is a score of 10 on Polity IV and thus Model 2 is restricted to countries that have an 11 or higher on Polity IV. We conduct robustness checks by running trials with a cutoff one point below (9) and one point above (11) the cutoff used in Model 2. (Please see Table A1 in the online appendix.)
[76] Tsebelis G, Nardi DJ (2016) A long constitution is a (positively) bad constitution: Evidence from OECD countries. British Journal of Political Science 46(2): 457–478. Crossref Google Scholar
[77] Dixon R, Landau D (2019) Constitutional end games: Making presidential term limits stick. Hastings Law Journal 71: 359–418. Google Scholar
[78] Kuenzi M, Lambright G (2001) Party system institutionalization in 30 African countries. Party Politics 7(4): 437–468. Crossref Google Scholar
[79] Cheeseman N (2018) Introduction: Understanding African politics: Bringing the state back in. In: Cheeseman N (ed) Institutions and Democracy in Africa: How the Rules of the Game Shape Political Developments. Cambridge: Cambridge University Press, pp. 1–38. Crossref Google Scholar
[80] Negretto GL (2012) Replacing and amending constitutions: The logic of constitutional change in Latin America. Law & Society Review 46(4): 749–779. Crossref Google Scholar
.