More than thirty years ago, the American Enterprise Institute (AEI) convened a set of politicians and lawyers who had participated in processes of constitution making. The resulting volume, Constitution Makers on Constitution Making, presented two authors from each of eight countries to consider issues of process and substance.Footnote 1 The approach of letting constitution makers speak for themselves, in a comparative framework, was novel and valuable. Constitution making is frequently idealized, and careful analysis of the perspectives of those closest to the process provided illuminating insights.
More than half of the world’s nations have rewritten constitutions since 1989, a year which marked an epochal change in history. These post–Cold War constitutions are qualitatively different than those that came before, containing more rights, a host of independent regulatory agencies, and more elaborate systems of accountability. Another change in constitution making came in the twenty-first century, in which new concerns about the balance between liberty and security have arisen. Recent years have seen new attention to issues of inequality, indigenous rights, resource use, and climate change. Security sector reform is another issue that has arisen with increasing frequency, as the boundary between peace treaties and constitution making becomes more blurred.
In parallel with the revival of constitutionalism in many countries, a new wave of academic work has emerged, providing a kind of renaissance of comparative constitutional studies (Hirschl Reference Hirschl2015). New theories and impressive developments in social science (in areas such as negotiation theory and behavioral psychology), along with more sophisticated empirical techniques, allow us to ask more sophisticated questions of constitution makers. This volume reflects an effort to leverage all these developments to answer the old question: how are constitutions actually made?
Under the auspices of The International Institute for Democracy and Electoral Assistance (International IDEA), we brought together a set of people who had been involved in constitutional drafting, to replicate the original project with a new set of case studies. We sought to take advantage of developments in scholarship and the real world since 1988, by addressing analytic problems in the field through the most recent cases. We wanted a set of cases that spanned geography and a range of constitution-making situations so as to capture the breadth of the field as it exists today. And we sought to understand how constitutions are made in reality: how agendas are formed, processes designed, and deadlocks broken. To use an American metaphor, we wanted the “inside baseball” of constitution making.
We made three important deviations from the process used by AEI in the development of the prior volume. Firstly, we brought a group of constitution makers together for four weeks at the Netherlands offices of International IDEA – rather than bringing them together for a single, two-day conference as was the case at AEI. This enabled a more in-depth exchange among the constitution makers, which improved the quality of the case studies through ongoing cross-learning amongst the authors. (Our Tunisia, Kenya, and Ecuador case studies were added later.) Second, we invited only a single author for each case, to maximize the comparative component of the inquiry.
Third, we asked the constitution makers to apply a specific analytic framework to their narratives to explain the process and results of the decision making related to constitution design choices. This both enabled a more comparative approach and allowed us to challenge, and update, the framework.
Our basic framework starts with treating constitution making as a bargaining process, in which multiple groups come together to try to resolve certain problems from the past, and to produce a constitutional text for the future that has broad agreement and social acceptance. Like all bargaining processes, the participants may have trouble reaching agreement, and may run into unexpected difficulties. Information – on the effects of different institutions, and on the intentions of other participants in the process – is not always available, and emotional factors may make bargains difficult even when information is relatively complete. Time pressures may work to force agreement or threaten to break it apart. And unexpected events can change perceptions and interests in the middle of the process.
We sought to explore the role of these challenges, and the bargaining and drafting techniques that allowed the parties to work around them. In this sense, we are “selecting on the dependent variable” by focusing only on cases where constitutions were actually completed. As we shall see, even these “success” stories feature many examples of unanticipated challenges and suboptimal outcomes.
Our starting point in thinking about process was the classic paper on constitution making by Professor Jon Elster, Forces and Mechanisms in the Constitution-Making Process (Reference Elster1995). Elster’s article is a masterful explanation of constitution making over time, noting that it has come in several waves. He draws on eighteenth-century thinkers to identify reasons, passions, and interests as major forces in constitution making. One of his useful points is that, despite our image of constitution makers as exercising the pouvoir constituent and speaking for a sovereign people, no constitution-making process is truly unconstrained. Instead, constitution making is subject to both upstream and downstream constraints. The former involves those that are set prior to constitution making, including the process of convening the body, any procedural rules or governing principles that are announced before it is established, and external forces that limit the substantive choices it can make. Peace agreements that precede constitution making are a frequent source of these constraints. Downstream constraints are those that flow from ratification, and to some extent from implementation as well. This simple framework of Elster’s provides a powerful and enduring lens through which to understand the dynamics of constitution making in many contexts.
At the same time, there are certain trends that have become apparent in constitution making in the post–Cold War world that require new analytic frames. Constitutions in this era have a different character than those that were produced in the earlier waves. In terms of process, there is the increasing role of the international community in constitution making, often as part of a broader intervention into a conflict. Constitution making is a key part of building new states (Wallis Reference Wallis2014). There is also the increasing penetration of international norms into national constitutional norms. Not unrelatedly, there has been a normative trend toward public participation in the process of constitution making – no longer is constitution making something that is done behind closed doors by small groups of elites. This trend has itself been promoted by the international community, as well as civil society, providing a twin constraint on national elites from both above and below.
Another trend has been the rise of constitutions that are produced in an iterated fashion over time, in which several documents are adopted in sequence, sometimes but not always labeled as interim or transitional. Andrew Arato (Reference Arato2016) calls this configuration of new features “post-sovereign constitution-making,” noting that it moves away from the paradigm of a single moment in which a sovereign people bind themselves. South Africa, which had both an interim constitution and a final one, as well as numerous political agreements along the way, is perhaps the paradigmatic example, but it is one that has been followed in many other countries. Of the cases examined in our volume, Burundi and Nepal mirror South Africa in their use of principles and interim constitutions, while the Kosovo process was also tightly bound by principles determined at the outset of the process.Footnote 2 Tunisia’s process featured a series of agreements as well. Kenya’s process followed an earlier, failed round of constitution making and so can be considered iterated as well.
Also with regard to process, Elster envisions ratification through public referendum as the archetypal downstream constraint, yet from the cases examined here only in Burundi, Ecuador, and Kenya did the constitution makers have to consider the hurdle of formal public approval. In Tunisia and South Africa, the processes involved a hybrid system in which there would be a public referendum if the elected constituent assembly could not agree on a draft with the required super majority. This incentivized drafters to complete the job themselves.
Lastly, unlike the two classic cases of France and the United States which are the focus of Elster’s analysis, many modern constitution-making bodies have included a growing number of women – 34.61% in Ecuador, 33% in Nepal, 28.1% in Tunisia, and 28% in Kosovo – and members of discrete minorities such as indigenous populations, who may be given designated seats. Many are selected through elections – as in our cases of Nepal, South Africa, Tunisia, and Ecuador. Elections should be considered part of the set of upstream constraints, as they provide mandates to the delegates.
Substantively, as mentioned above, certain trends are also apparent in constitutional design. These include: the increasing deployment of federalism and other means of spatial decentralization of power to resolve internal violent conflict; the rise of “fourth branch” institutions that can monitor the performance of government, such as National Human Rights Institutions (NHRIs) and counter-corruption commissions; the growing powers of constitutional courts; the increasing number of rights clauses in national constitutions; new attention to the environment; and a rise in concerns of identity and inclusion, for ethnicity, gender, and other dimensions of social difference.
To examine these issues, we looked at a set of seven cases: Burundi, Ecuador, Kenya, Kosovo, Nepal, South Africa, and Tunisia.Footnote 3 These are very different environments for constitution making, and our criteria in selecting these cases from a broader set of applicants reflected a desire for diversity in terms of region, system of government, and type of process. Several of these are countries recovering from violent conflict and war, a factor which tends to encourage international attention. Constitution making in our era is transnationally embedded, and our cases reflect a range of levels of engagement with the international community. In Burundi and Kosovo, international involvement was intensive and sustained, and constitution making can be understood as emerging rather directly out of a peace agreement among warring parties. In South Africa, Ecuador, Kenya, and Nepal, there was a good deal of attention from outside, but also a good deal of constitutional expertise in the country, and the primary shape of the deal was locally determined. Tunisia’s dynamic was local revolution, and the process was locally managed.
Gargarella (Reference Gargarella, Ginsburg and Huq2016) following Alberdi (Reference Alberdi1981 [1852]) argues that constitutions should be assessed first by their ability to address their “central dramas.” Each of these cases had a very different drama, though there are some interesting connections and parallels across the cases. Kosovo involved the drama of securing national independence in a highly fraught context, in which memories of the wars that accompanied the breakup of Yugoslavia remained fresh. South Africa was also an instance of what the judge and activist Albie Sachs has called nation building in the aftermath of decades of apartheid and violent conflict. Burundi, too, involved memories of ethnic conflict, including multiple instances of genocide in the country’s history. Nepal’s “People’s War,” led by the Maoists, had an ethnic as well as a class dimension, and had the goal of producing an inclusive federal constitutional order. Ecuador’s efforts involved regional and ethnic conflict. Kenya’s central drama was the struggle for governmental integrity, but it also occurred in the aftermath of civil conflict, as ethnic-based violence had followed the prior election. And Tunisia’s constitution making followed its popular revolution that triggered the Arab Uprising throughout the Middle East. Thus, a common motive triggering constitution making is an effort to find a more effective, and in some cases inclusive, model for governance. It was also apparent from the cases that the definition of the central drama might not be agreed at all levels – for example in Tunisia much of the debate within the Constituent Assembly was focused on the identity of Tunisia as secular or Islamist, as well as the character of the political system, whereas for the younger revolutionaries the transition was more about social justice and human dignity.
These dramas, to which constitution making responds, frame the political process of bargaining under constraints. Constitution making, as Gabriel Negretto (Reference Negretto2013) has noted, involves mixed games of cooperation and competition. Parties may have a joint interest in resolving certain core challenges, but may differ in terms of the particular solutions, which often have distributive consequences. It is this process of sorting out the costs and benefits, of creating new institutional structures, and of inspiring the public to support the outcome, that successful constitution making must grapple with. Beliefs, assets, and constraints come into play; deadlines and timelines matter, as does the sequence in which issues are tackled. We seek to address these issues of time, information, and constraint in the bargaining process, and to deepen our understanding of the roles of public participation, elites, and the international community in constitution making.
The remainder of this introduction is organized as follows. We first focus on the issue of assets and constraints that designers bring to the table; then turn to the role of uncertainty and beliefs in the process. We then examine lessons from the cases in terms of process design, international involvement, public participation, and the role of courts. We subsequently turn to lessons from the cases for substance. We examine the production process of two aspects of constitutions, the preambles and the transitional provisions, which typically form the bookends of the text. How these particular parts are drafted provides an interesting insight into the way constitution makers proceed through their task. We conclude with reflections on the range of lessons that can be drawn from our cases.
I. Assets and Constraints
Constitution making is, in our conception, the process by which a formal constitutional document is produced. The process begins with the decision to draft a new constitution and ends with the promulgation and entry into force of the resulting document. It includes numerous internal stages, such as gathering ideas, negotiating, drafting, reviewing, and perhaps seeking public approval. This process can take anywhere from a few months to many years.
Clearly, this description of the formal process is inadequate in that constitution making is embedded in a larger set of historical struggles, typically beginning long before formal constitution making is launched and ending well after formal promulgation. These background dynamics determine the relevant upstream and downstream constraints, in Elster’s terminology. In some cases, the key constitutional decisions may be taken earlier, in a peace negotiation or other context. For example, in Kosovo, talks held in Vienna under the auspices of mediator Martti Ahtisaari resolved some of the major issues for a future constitution, well before it was clear that such a constitution would indeed be produced. And one nominal downstream constraint – the formal adoption of the constitution by the elected assembly – proved to be meaningless relative to the need for international approval thereof. Thus, the notion of constraints is a fluid one and requires case-by-case analysis to determine what limitations actually are present on the process.
In South Africa, one can trace the struggle for a democratic constitution back many decades. In the immediate pre-constitutional period, political prisoners were released, and initial negotiations began. Upstream political agreements made during this period had important implications for the process of future constitutional negotiations. For example, it was agreed at the CODESA I talks, at an early stage of the political transition, that decisions in the South African transition should proceed on the basis of “sufficient consensus,” an undefined term that was interpreted to mean that only the two main players would have a veto. Later, this led to agreement on the rule for adoption of the final constitution: it would have to be agreed by a two-thirds majority, and if that could not be reached, a draft with the support of a majority of members could be sent to referendum.
Given the iterative nature of modern constitution-making processes, constraints also evolve through the course of the same process. Nepal is perhaps the clearest example of this, involving, as it did, two consecutive constituent assemblies with vastly different compositions. Initial upstream constraints on the first process were not airtight, but they did eventually lead to its failure, along with the establishment of a new set of rules for a second Constituent Assembly that accomplished its goal.
Another set of constraints comes from the structure of the cleavages at issue. South Africa was essentially a bilateral negotiation, pitting the white apartheid forces against the African National Congress, but there were other parties whose interests had to be taken into account, most visibly the Inkatha Freedom Party. Burundi’s conflict was essentially a bilateral ethnic one, pitting Hutu and Tutsi forces against each other. Constitution making in Kenya featured a number of different elite groups, whose coalitions shifted over time. Ecuador’s exercise was one of intense regional conflict, while in Kosovo, there was internal unity at the early stage but also some bilateral personality-based politics. The principal societal cleavage – relating to the Serb minority – had been removed from the constitutional negotiations by the preceding Ahtisaari principles which dictated a series of minority rights and protections with which the constitution makers were enjoined to comply. The most complex of our cases was Nepal, which is an utterly diverse country with lots of cleavages. It is tempting, but in our view mistaken, to attribute to this diversity the failure of Nepal’s constitution-making process to produce a complete settlement. After all, we know of other cases of unilateral constitution making – Georgia is an example, though outside the scope of our volume – that ended up producing an unstable solution because of the surprising results of the post-constitutional elections. Instead, we would argue that multisided bargaining is more difficult but that it was specific choices that led to Nepal’s instability.
Constraints can also be indirectly articulated by international financial organizations. Tunisia offers the best example of such an external constraint, as the country saw its credit rating decline as the constitutional drafting process went on. These external decisions placed further time pressure on the MPs to conclude a deal. Credit rating agencies, along with the International Monetary Fund and the World Bank, may prioritize economic concerns over long term democracy building.
The background or upstream iterations of bargaining also determine the assets that constitution makers bring to the table. If constraints are limits that shape the process, assets are conditions or features of the environment that facilitate the process of constitution making. In some cases, these may be a set of accepted commitments and ideas about how future governance ought to work; in other cases, the assets might include political party structures or historical leaders. For example, the presence of a unifying leader can be a major asset. The presence of George Washington in Philadelphia, a man who everyone agreed would serve as the first president of the country, helped to channel disagreements on other issues.
In our cases, the unquestioned position of a Nelson Mandela helped to channel conflict and overcome constraints in South Africa. Ebrahim reports that, in South Africa, Mandela would attend meetings but rarely speak. While not involved in the decision-making meetings, Mandela’s presence in the background provided reassurances for all sides. The presence of a Mandela – or in other contexts, a George Washington or Jawaharlal Nehru – is a great advantage but not one that can be engineered. (Interestingly, Mandela also appeared in Burundi as an external mediator, pushing the parties toward agreement.)
Beyond resolving conflicts, the inclusion of eminent figures can also help to add weight to the constitution-making body, and as a result increased legitimacy to its proposals. In the United States in 1787, the inclusion of an ailing Benjamin Franklin helped to strengthen the authority of the Philadelphia Convention.
The availability of local political and legal expertise is also an important asset. South Africa’s African National Congress had a good deal of legal and political talent, developed over many decades. It also was able to channel the various strands of the anti-apartheid movement into a single voice, more or less, securing a better deal than had it been fractured.
Locations can be assets too. After the fall of communism but before the Yugoslavian breakup, Kosovo leaders gathered in the historically important town of Kaçanik to adopt a new constitution for a republic, within the framework of Yugoslavia. The location helped to contribute to a narrative. In another context, the 2007 Constitution of Bolivia was to be drafted in Sucre, a relatively neutral site between two competing regions of the country (Landau Reference Landau2013: 953).Footnote 4 The earlier round of Kenyan constitutional negotiations had taken place at Bomas, outside the main center of Nairobi. For negotiations, locations have symbolic meaning, which makes some better than others. The Vienna talks on Kosovo could not have been held in Berlin, Washington, or Moscow, for each of these places would be viewed as partisan. Instead, having a Finnish negotiator lead talks in Vienna projected a sense of neutrality for the partisans.
The state of the economy may also be an important factor as either asset or constraint. In Tunisia, the deteriorating economy and high unemployment placed pressure on the constitution makers to complete their task.
In short, some assets can be engineered, but others are simply a product of the situation of constitution making. Constraints are typically taken as given, but sometimes result from earlier rounds of negotiation and interaction that shape the constitutional moment.
II. The Role of Uncertainty and Beliefs
Constitution making is often conceived of as a Rawlsian process, in which institutional designers are picking institutions in ignorance of the position they will hold in post constitutional governance. Their beliefs about the outcomes are important in incentivizing the choices they make.
The cases we have examined suggest that there are two common reasons that things may not go quite as planned: random shocks and electoral surprises. Random shocks are truly unpredictable events that occur in the course of constitution making: The 2015 earthquake in Nepal is a paradigm example. That event strengthened the hand of the majority to push through a deal on the constitution, though as Dev writes, it did not actually lead to a more accommodating bargain. And in South Africa, the assassination of Chris Hani in April 1993 was a significant event that surprised everyone. The shock changed beliefs about the probabilities of certain events, making the specter of civil war more salient for all, and this threat was leveraged to reach agreement on the timing of elections. In both cases, the shocks created crises which sped up the timeline of constitution making. On the other hand, the assassination of Tunisian Constituent Assembly member Mohamed Brahmi in July 2013 froze the process for months and threatened to derail it entirely.
Electoral surprises, of course, are taken into account in the standard framework. But what is perhaps interesting is how often politicians involved in constitution making seem to miscalculate the electoral results. In Kenya, Raila Odinga pushed for a new constitution, perhaps hoping to take the presidency, but the International Criminal Court’s decision to indict Uhuru Kenyatta and William Ruto on the eve of the first post-constitution election led to a backlash that swept them into office. While we do not analyze the case in this volume, constitution making in Georgia was pursued under the very strong conviction that Mikhail Saakashvili’s party would win the post-constitution-making election; it did not, despite the fact that there was no major challenger in sight beforehand. The sudden political emergence of Bidzina Ivanishvili, who formed his Georgian Dream Coalition six months before the October 2012 election, proved an unexpected surprise. Saakashvili’s idea of strengthening the Parliament and the state apparatus proved to be unrealizable in an era of informal political domination by a single figure.
In Burundi, constitution making in the interim period of 2001–2005 proceeded during a period when two major political forces were to alternate interim presidencies of eighteen months each. Neither holder would be allowed to contest the first post-constitutional election. As constitutional decisions had to be taken as to how the president would be selected under the final arrangement, Jean Minani, a leader of the FRODEBU party, decided not to take the interim presidential term, as he thought that the Parliament would be a better base in which to take the permanent presidency after constitutional adoption and the interim president would not be eligible to run for presidential office at the end of the interim period. However, he did not anticipate that the rival CNDD-FDD party would contest the elections, and surely did not expect that they would win. When CNDD-FDD won the elections under a constitution they had no role in designing, they had little investment in implementing it, and President Pierre Nkurunziza went on to serve three consecutive terms in office, with grave human rights abuses taking place.
Electoral uncertainty also played a role in the Nepal case, though in a less unpredictable way. When the first Constituent Assembly failed to complete its work, the Supreme Court ordered it disbanded, leading eventually to new elections. From the perspective of the sitting politicians, the court case was a kind of random shock that they did not expect. But the subsequent elections, which produced a political configuration completely inverse to the first Constituent Assembly, was an example of uncertainty shaping the subsequent agreement. In turn, the second Constituent Assembly was guaranteed to remain as a legislature in the post-constitutional period, reducing uncertainty for individual politicians and perhaps facilitating the ultimate agreement.
The bargaining process can also lead to parties switching positions, as they learn from each other or from new information. Proposals to turn Kenya’s system into a parliamentary one have surfaced from time to time but never overcome the logic of presidentialism. In Nepal, the Communist Party Nepal – Maoist (CPM(M)) abandoned its position on presidentialism. The fluidity of interests, and the repeated failure of political actors to predict post-constitutional politics with accuracy, is an important theme that deserves explicit inclusion in a neo-Elsterian framework. Despite pervasive uncertainty, there seems to be a good deal of optimism bias among actors drafting behind the veil of ignorance.
III. Lessons from the Cases: Process Design
Our cases tell us a good deal of how constitutional process can be used to facilitate agreement, resolve deadlocks, and either help or hinder the constitution makers in achieving their goals. We address several different issues: the design of the forum, the sequencing of issues, the role of deadlines, secrecy and publicity, and mechanisms for breaking deadlock.
A. Plenaries, Back Channels and Forum Design
As a formal matter, constitution making is often developed through constituent assemblies, either specially elected for that purpose or doubling as a sitting legislature. In South Africa, the Constitutional Assembly was both houses of Parliament, sitting together in plenary. In Nepal and Tunisia, the Constituent Assembly was elected as such and then doubled as the country’s legislature. In other cases, the legislature itself may produce the constitution. In the unusual case of Kosovo, a constitutional commission did the formal drafting, on the basis of a working group that had done much of the work already; the Assembly played only a minor role in giving final approval to what was largely a fait accompli. The process was similar in Kenya, although in that case a parliamentary committee of political leaders had a significant role in shaping the draft developed by the expert commission, and the draft approved by Parliament was still subject to popular referendum.
There is something of a paradox in Constituent Assembly design. A larger body can be important to guarantee inclusion; but at the same time, larger bodies may be challenging to manage, and may make coming to agreement more difficult. In virtually every case, therefore, there are smaller groups of major political actors that make the major and contentious decisions. South Africa had, within the Constitutional Assembly, a ten-member management committee, and also frequently formed an informal “channel bilateral” to hammer out specific compromises among the key negotiators. Much of the key decision making occurred in these smaller, informal meetings while the Constitutional Assembly as a whole met only a few times in plenary. Kosovo had a “Unity Team” of political representatives that stood over a more technical body preparing the constitutional draft. This allowed good coordination between technical and political levels. In Tunisia, an unofficial “Consensus Committee” included the president and vice presidents of the Constituent Assembly, the rapporteur, and two members of each political party represented in the Assembly. This crucial committee was tasked to reach compromise on the most complicated issues without involving the entire Assembly.
Instead of focusing on the Constituent Assembly as the ultimate body, therefore, it might be better to think of constitution making as involving a series of four concentric circles. At the broadest, constitution making is done by “the people themselves” (Kramer Reference Kramer2002) in whose name the exercise proceeds, but of course it is only a useful fiction that the people actually produce the document that binds them. More narrowly there is the broad set of elites that run the country, a group that varies in size across countries (Bueno de Mesquita et al., Reference Bueno de Mesquita, Smith, Siverson and Morrow2003). These elites include elected or designated representatives from the broader circle of the people, but would also include academics, judges, traditional leaders, and many others. Within this set of elites, a smaller set are selected to serve on the official constitution-making body, whether it be a Constitutional Commission, Constituent Assembly, or sitting legislature. This group will be supported by a staff and by outside experts who contribute to the process, and together these groups are the actual writers of the constitution. But within this relatively narrow circle (and perhaps even including leaders who are formally outside it), there will be a smaller group of high-level decision makers who have the power and authority to commit to making decisions on behalf of their relevant constituencies. This core group does not, by its nature, actually do the drafting of constitutions; instead it serves as the group whose ideas and decisions are translated, sometimes imperfectly, into the final document. This group is, of course, constrained by the other circles in various ways.
Procedural constraints operate within every one of these circles, but among the most critical will be the decision-making rule in the formal negotiation body. Oftentimes, this is something broader than simple majority, in light of the symbolic importance of constitutions. In South Africa, as mentioned above, decision rule of “sufficient consensus” meant something more than majority but shy of unanimity – essentially, it meant the agreement of the two key players, the African National Congress (ANC) and the National Party (NP) but without dominating the core interests of smaller parties. The ANC had taken the decision to proceed as much as possible through consensus, so as to mark the way forward to a new South Africa characterized by inclusion rather than domination. But at the same time, unanimity in any negotiation process threatens to incentivize holdouts and brinksmanship. The sufficient consensus formula included the two main players, but not the minor ones. While efforts were always undertaken to bring the minor players on board, they were not given a veto. A similar approach was taken in Arusha in the negotiations over Burundi’s settlement: bringing in the main players without giving smaller ones a veto.
Whatever the formal plenary body, it is typically divided into thematic committees to accomplish the task. These are usually aligned with different substantive parts of the constitution. In Nepal, interestingly, the second Constituent Assembly divided itself into functional committees, including a “Constitutional Dialogue and Political Consultation Committee” (on which Dev served), to discuss core contentious issues and was at the heart of the process.
The plenary body and/or the committees are also supported in some cases by technical or staff committees to advice and draft. These actors can sometimes be powerful: one story that emerged in one of our case studies suggested that staff responsible for typing up the decisions from daily meetings would sometimes modify the record, forcing the political negotiators to monitor staff closely. Constitutions may be “higher law” but they are subject to the same human foibles found in every process of governance, in which there is sometimes slippage between supervisor and staff.
B. Role of Deadlines and Timing
Time is a major constraint in constitution making, but time can also be leveraged into an asset. As time grows short, deadlines can force parties into agreement so that they cannot be blamed for the failure of the process. The speedy timeline pushed by the ANC in South Africa, relative to the more extended period that had been demanded by the National Party, meant that neither party wanted to be seen as a spoiler in the process.
Burundi’s case was unusual in that it emerged as part of a set of international negotiations that were initially open-ended. The process involved a set of committees that dealt with particular issues, meeting in parallel without time limits. Eventually, time limits were introduced to facilitate a successful conclusion in 2000. This led to an interim constitution and government that was to include two alternating periods of 18 months for each of the main negotiating forces to hold the presidency. (This alternation mechanism was effectively implemented, but as described above in the discussion of uncertainty, had significant downstream effects when neither party won a subsequent election.)
In the case of Kosovo, constitution making was bounded by the need to secure support for independence, leading to a very narrow window for formal drafting and adoption. The process had to begin without knowing whether a Coordinated Declaration of Independence would even take place, much less when, but drafters did know that the Declaration would trigger a 120 day period in which the constitution and much other legislation had to be adopted. This served to focus the drafting into the earlier, secret phase, and may have forced political rivals to set aside conflicts in the pursuit of their collective interests in speed. The combination of secrecy, a short timeline, and very high stakes in which all parties shared certain interests (namely, securing independence) served to facilitate an efficient drafting process.
Nepal’s Constituent Assembly initially had a two-year window to complete the task; it failed to abide by this deadline and granted itself two extensions, until the Supreme Court intervened to reject a third. New elections were held, but the second Constituent Assembly was equally laconic. Deadlines in Nepal seem to be flexible, but this has a negative impact on the process because parties seek to outlast each other in refusing to compromise.
Tunisia exemplifies the risk of leaving the mandate open-ended. After the dictator Ben-Ali fled the country, a “Higher Instance” was put together with law experts and members of the oppositions in order to create a post-revolution roadmap and prepare the necessary laws for the election for the National Constituent Assembly (NCA). Despite being an unelected body, the Higher Instance decided to bind the NCA members with a moral commitment to draft the constitution within a year. However, once elected, the NCA decided to define the end of its own mandate with the “adoption of a constitution.” This lack of timeline provided ammunition to the counter revolution, which led to a crisis of legitimacy. As a result, a year and a half after its election, the NCA faced campaigns for its dissolution.
In Kenya, the statutory framework established a tight timeline with twelve discrete steps for the Committee of Experts to complete. All steps were completed within the overall timeline, albeit with an extension owing to a delay of three months in establishing and appointing the Committee.
In short, time matters for constitution making. An open-ended process will not likely succeed; at the same time, process designers would be wise to recognize that drafters and negotiators need time to accomplish their work. A noteworthy recent case was that of Libya, which initially allowed a period of only four months for constitution making. When that deadline failed, it led to a seemingly unending process of drafting.
In many instances, constitution-making processes last longer than intended or expected at the outset. Tunisia took two years; Nepal eight. One advantage of extended time periods is that the negotiators can build up personal relations and trust, despite partisan differences. But external forces are likely to impede efforts to extend the timeline too long.
C. Sequence of Issues and Agendas
Negotiators have a good deal of discretion in deciding which issues to take on in which order, and these decisions can be consequential. Patterns of agreement or disagreement unfold over time; if an early deadlock persists it can undermine the entire process. There has, to date, been little attention in the literature on these choices and their consequences in practice, even though a whole branch of constitutional economics, beginning with Arrow (Reference Arrow1951) has examined them in theory.
In South Africa, the Constituent Assembly Secretariat pursued a practice of having a single draft during the entire process of negotiation, which necessitated bracketing controversial issues. They developed the concept of a “parking lot” of issues that would be decided later off. (As Ebrahim noted in our discussion, this had the virtue of lowering the stakes of these issues, when they were eventually tackled late in the process.) The issues included in the parking lot tended to be important enough to be contentious, but not so important as to be deal breakers. South Africa thus provides an example of establishing deadlines, meeting them, and completing the task in an effective manner. The success of the South Africa process in observing its pre-set timeline can in part be attributed to the two-stage process and the fact that many of the most contentious issues were decided in the interim constitution before the Constitutional Assembly was elected. In Nepal, however, while a similar two-stage process was followed the most contentious issues were left to be decided by the Constituent Assembly which – as discussed above – failed to meet several deadlines, resulting in numerous extensions to the process, including the dissolution of the first Constituent Assembly and fresh elections.
In Kosovo, the technical subcommittees that were doing the drafting returned to their political principals for guidance but did so through a single multiple-choice questionnaire that required unified answers from the political group. This sequence meant that the technical drafters had the effect of setting the agenda, by implication narrowing it. But it also presented challenges as they had to integrate a set of decisions made by the political principals.
In Tunisia, Constituent Assembly members decided to start jotting down ideas, and discussed them before even starting the drafting process. The publicly articulated objectives of the revolution were guiding the process, as was each political party’s platform and history.
The Kenyan process of allowing a Committee of Experts to produce the first draft meant that there was little direct political intervention in the sequencing of issues and agendas. Furthermore, the Bomas Draft that had been produced in a prior round of constitution making had articulated many of the ideas for constitutional reform, so the Committee had a template to start with. But when the process returned to the Parliament in the form of a Select Committee, it was able to modify some of the details that had been agreed.
A final point that emerges from our cases, constitution making is often focused on the central dramas, but it is sometimes the smaller issues that become barriers, even if not anticipated at the outset of the process. Sequencing issues cannot always be managed in a technocratic manner.
D. Secrecy and Publicity
In Kosovo, the process was by necessity initiated in secret because the international community had not fully agreed to independence. A draft constitution had to be fairly far along so that it could be quickly adopted in the event that occurred. As Caka notes, time went on and a decision to adopt a Coordinated Declaration of Independence was finally made, but the drafting process was not formally initiated until the Declaration on February 17, 2008. When the draft, already developed over the course over the previous year, was finally put on the web, public input began.
In Burundi, the formal plenary processes were proceeding in parallel with decisions being made privately by leaders of the two largest blocs. These decisions were then explained to the political parties, a process that led to occasional problems, for example when the speaker of the National Assembly said the process was almost over, when the actual plenary members did not know this.
In South Africa, Tunisia and Nepal, the Constituent Assembly sessions were open to the public, and in the case of Nepal were even televised.Footnote 5 However, in all three cases, critical decisions were often made behind closed doors within a very small circle of participants. In the case of South Africa, this was not problematic, as the key negotiators which formed the “channel-bilateral” represented both major sides of the bargain, and as individuals they commanded a high degree of legitimacy and authority among the broader circles of negotiating actors. Similarly, in Tunisia, specific language that could not be agreed upon by members of the constitutional committee were further discussed in the evening between the major political leaders. Oftentimes, the next morning, the rapporteur would bring a new sentence and the parties would agree automatically without further debates. Not all members of the committee knew about these secret meetings, which created some frustration among them. But it served to facilitate the process.
However, in Nepal, a significant initial trigger for resistance by minority parties to the proposed constitution was the nature in which it was finalized in secret by a small, noninclusive group representing only the majority parties. The key difference seems to be whether secrecy is being used to facilitate bargaining, or to avoid it.
In Kenya, deliberations within the Committee of Experts and the Parliamentary Select Committee were generally closed to the public. Indeed, there remains a fair amount of mystery surrounding the Parliamentary Select Committee (PSC) retreat at which some major issues were agreed upon. However, this secrecy was counterbalanced with a high degree of transparency over the process as a whole, including the publication of an initial draft for public consultation.
E. Techniques for Breaking Deadlocks
Because it involves a mixed game of conflict and cooperation (Negretto Reference Negretto2013), constitution making rarely proceeds without serious conflicts. These can be the result of genuine distributive conflicts, but they can also result from bargaining failures. A party can “hold out” for a better deal but find that its interlocutor is doing the same thing. No one has an incentive to reveal its true “bottom line” if they believe they can get a better deal.
There are, of course, myriad techniques to try to resolve bargaining failures among constitution makers. One solution when negotiations are deadlocked was used in South Africa – the Secretariat took the negotiators to a resort location, far from the pressures of their constituents, to break the ice, talk informally, and make key decisions in a less pressured atmosphere. This same technique was used with parliamentarians in the Kenya process of 2009–2010.
Substantive provisions can also serve to help resolve deadlocks by expanding the benefits of the constitution. Among the motivations driving constitution makers, personal and partisan interests are certainly present: Where will I be situated in the new constitutional order? How will these rules affect my party? In some cases, creating more government bodies can help to resolve deadlocks because it creates more possible positions for the negotiators themselves. In the case of Burundi, a small party led by a relative of the former king demanded political representation, and the negotiators included a provision that a referendum on restoring the monarchy could be held in the future (though it has never been held). This holding out of a potential future benefit served to ameliorate the concerns of one party.
Another technique is “deciding not to decide” (Dixon and Ginsburg 2012). Postponing certain issues to the end of the process is a strategy that was used to great effect in South Africa, in the form of the “parking lot” that has already been mentioned. In other cases, decisions were taken early to leave the substantive choice for the post-constitutional order: In South Africa, the death penalty was an example here, as was, to some extent judicial restructuring. (A classic example from the literature was the decision in India to defer the issue of a unified civil code until after constitution making, a decision that shaped the early years of constitutional operation.)
In Kosovo, when marriage equality became a controversial issue, the drafters adopted a formula that postponed the decision to law but mentioned that the right applied to “everyone.” This gave both proponents and opponents of marriage equality a sense of achievement. Such abstraction (which Sunstein 1995 calls an incompletely theorized agreement) is a form of deciding not to decide everything. Tunisia’s first two articles define the country as a civil state while also noting that the religion is Islam, leaving the precise relationships to be worked out later. In South Africa, the formulation of socioeconomic rights recognized that they would have to be implemented over time, and so were not defined with great specificity. And in Burundi, the Senate was given a role in monitoring the ethnic integration of the army, helping to serve as a guarantor that the goal would be met.
Similarly, in Kenya, the draft constitution had a provision that stated that life began at conception, and that abortion was not permitted, but went on to say it would be allowed if “in the opinion of a trained health professional, there is need for emergency treatment, or the life or health of the mother is in danger, or if permitted by any other written law.” This meant that in the end, there was discretion on the part of the legislature to deal with this controversial and contentious issue.
While instances of abstraction to resolve deadlocks are sometimes effective, they are not always so. In Burundi, the right to property was controversial because prior rounds of ethnic cleansing had displaced people. These issues continue to linger today.
The theory of deciding not to decide suggests that it is an advisable technique when it is unclear whether the negotiators have full information to make the right decision. This helps shed light on a bargaining failure in the context of Nepal. As Dev describes in his chapter, the issues of identity and federalism were absolutely core, and there was a good deal of distrust among the parties. This led his Madhesi constituency to demand to more specificity in the text of the constitution, as they believed that their interests would not materialize if details were postponed until later. At the same time, and for the same reasons, the amendment rule was set very high: To change the boundaries of provinces requires a two-thirds majority in Parliament, as well as the agreement of local populations, and a majority of total provinces in the country. This meant that, once adopted, it would be hard to change the borders of the internal units. And yet information as to what would constitute coherent and viable units was lacking, since Nepal had never been a federal country and was creating the subunits from scratch. This had the effect of raising the stakes of the initial drawing of boundaries to be very high indeed, potentially freezing in place a suboptimal arrangement. The effect has been the failure to resolve the core issues.
IV. Lessons from the Cases: International Involvement and Public Participation
A. International Involvement
International involvement in constitution making has become both more intensive and extensive in recent years: it is more common and sometimes quite deep (Al-Ali and Dann 2005; Al-Ali Reference Al-Ali, Dixon and Ginsburg2012; UN Secretary General 2009). This fact raises interesting normative questions about sovereignty, as well as positive questions about how decisions are actually made.
The Kosovo case in particular is one in which international involvement was a necessary condition, given that the country’s successful creation was utterly dependent on recognition from powerful external actors. The very design of the process, and final approval, were in the hands of the international community, although there was some consultation with Kosovars to be sure.
As Caka notes in her case study, Kosovo’s story is embedded in the larger context of the Yugoslav wars of the 1990s. Kosovo’s independence struggle, though foreshadowed by its argument for a constitutional status within the Federal Republic of Yugoslavia, was really launched with UN Resolution 1244 and the establishment of the United Nations Interim Administration in Kosovo (UNMIK) in 1999. Through UNMIK, negotiations as to the future status of the territory intensified around 2004 with the appointment of former President Martti Ahtisaari as a mediator.
The international community, already deeply enmeshed in the region through the Bosnia-Herzegovina settlement, set itself the task of safeguarding the interests of the Serbs of Kosovo, who refused to participate in the constitution-making process. It thus served as a sort of trustee for part of the population of Kosovo. This affected the text of the constitution in several ways, downplaying the majority Albanian ethnicity, and arguably securing for Kosovo’s Serbs a better deal than they would have been able to obtain on their own. In this sense, the upstream and downstream constraints offered by the process were significant and consequential. And the international strategy meant that the Serb minority paid little price for holding out.
At the same time, international constraints improved relations among the local Kosovar politicians, who had to maintain unity in order to steer the country to independence. Caka coins the phrase “passion management” to capture how international involvement dampened the emergence of any maximalist demands, illustrating one of Elster’s points about the dangers of passions.
The framework imposed by the Ahtisaari plan both set out constraints but also left key questions about the structure of government unresolved and free for local drafters to decide. Major issues of legislative–executive relations were left to local determination. Thus, relative to Bosnia-Herzegovina, the constitution of which was included as an Annex to the internationally brokered Dayton Agreement, local involvement was genuine. This allowed for some ability to soften the demands of the upstream constraints: The decentralization provisions demanded by the Ahtisaari plan, for example, were left to law rather than included in detail in the constitutional text.
The downstream constraints of obtaining certification from the International Civilian Representative (ICR) were very significant. Yet even they were not dispositive. The ICR wanted to ensure that the preamble characterized the society as “multiethnic” (hence minimizing the Albanian identity of the majority), but at the end of the process the Constitutional Commission modified his suggestion and specified that the multiethnic nation consisted of Albanian and other communities, effectively calling the bluff of the international representative at the final stage.
In Burundi, too, the international community played a central role in determining the substance of the bargain, and formal constitution making was more of a pro forma exercise to bring this bargain into force. The Arusha Accords, signed in 2000 after intensive negotiations, sought to end an extended period of ethnic conflict and violence between Hutu and Tutsi. While the country had suffered cycles of ethnic violence and genocide since independence, the events in the mid-1990s in neighboring Rwanda focused the attention of the international community on the country. The presence of high-level mediators, including Julius Nyerere, Nelson Mandela, and Thabo Mbeki, was critical in bringing the rival parties to the table.
Kenya was an interesting case in which international pressure at the highest level was brought to bear after election violence. The content of constitution making was left to locals, but the designers of the process included three foreigners to sit on the Committee of Experts preparing the draft. As Murray (who was one of the three) points out in her chapter, this provided for some technical input and allowed a perception of neutrality.
International involvement was much less extensive in Nepal. Although the two large neighbors, China and India, have their interests, for the most part they let constitution making proceed without pressure. Foreign nongovernmental organizations (NGOs) and advisors, however, did push for particular substantive positions, as they frequently do in the current era in many countries. And the UN was present providing support. The Western donors played some role in pushing the parties to final agreement after the 2015 earthquake, as a certain amount of “donor fatigue” set in. But altogether, foreign involvement did not determine the key substantive decisions. And the country was quite open to foreign advice. In Ecuador, there were some foreign advisors from a Spanish university, but their impact is difficult to determine.
In South Africa, outsiders played only a minimal direct role, but international influence could be felt in other ways. Firstly, the political castigation of the apartheid regime by most Western powers, coupled with economic sanctions, resulted in steady economic decline, which forced the ruling party’s hand in beginning negotiations on the political transition. Secondly, the need to re-establish the country as a legitimate member of the international community of liberal democracies was influential on the content of the constitution, as is made explicit in the goals articulated in the preamble. In this regard, the need for international legitimacy proved to be both an upstream and downstream constraint.
In Tunisia, Western embassies discreetly pressured the Constituent Assembly to drop mentions of “solidarity for Palestine” and the “struggle against Zionism” in the text, as well as an article excluding members of the old regime from seeking future office. Germany was especially direct, threatening to withdraw aid to Tunisia should the constitution include the exclusion of the old regime or mention “the struggle against Zionism.” These pressures proved to be counterproductive, as they were viewed as an infringement on Tunisia’s sovereignty and actually led many drafters to dig in their heels on these issues. The explicit mention of Zionism was eventually dropped in favor of a statement of solidarity with the Palestinian liberation movement in the preamble, but Mabrouka argues that it is difficult to gauge foreign influence on this final result.
The international community can play an important role as monitors of the agreement as well. At several points in the interim period before constitutional adoption, international mediators intervened to remind parties of their commitments. In Tunisia, the Venice Commission and other international actors provided some technical advice to the drafting committee. In Burundi, the international mediators were absolutely critical in producing an agreement. The most significant international role was in Kosovo, where the ICR played a supervisory role after the constitution was adopted. In addition, the presence of an international audience is a form of downstream constraint, especially so in Kosovo.
B. Public Participation
Constitution-making processes can provide an opportunity to engage the population in a deliberative project that can have important salutary effects on the polity. For this reason, it has become a norm for the drafting of new constitutions in restructuring states that the population be widely consulted, and typically also have the role of approving the constitution through referendum. The constitution-making project can help to cement the nation’s sense of itself.
The trend toward greater and greater public participation in constitution-making processes takes many forms. The typical model is a public referendum at the end of the process to legitimate a new bargain. This mode of participation is neither deliberative nor deep. A more robust form asks citizens for input at the idea-generation stage, and recent work has suggested that this early involvement is more consequential (Eisenstadt et al Reference Eisenstadt, Carl LeVan and Maboudi2017). Jon Elster, in his classic article, asserts that the ideal shape of public participation would resemble an hourglass. It should be very wide at the beginning of the process, in the stages of idea generation, but then would narrow, when hard political negotiations must be conducted without the pressures of public transparency. At the end of the process, the public could be involved again in adopting and ratifying the constitution. This normative idea of Elster’s has been very influential and captures intuitions about how constitutions actually work.
Empirical evidence is mixed as to the benefits of public participation. Abrak Saati (Reference Saati2015) goes so far as to identify a “Participation Myth.” While many international organizations laud the benefit of public participation, she finds in case study evidence that it does not meet the claims of its proponents, specifically with regards to increasing levels of democracy. In an early study, Blount and others (2009) found that constitutional participation had little impact on the final products. But in a recent large-n study, Todd Eisenstadt and co-authors (Reference Eisenstadt, Carl LeVan and Maboudi2017) find that participation is most effective when it occurs early in the constitution-making process. A mere referendum at the end does little, but early participation increases the democratic quality and performance of the final constitution.
Our cases do not fit easily into either camp. In South Africa, there were efforts to inform the public of the ongoing work on the constitution, and of course the ANC had the foundation of decades of experience in encouraging local participation. The 1955 Freedom Charter, for example, was adopted after a broad public participation process, with more than 50,000 volunteers sent out to cities, villages and remote areas in the country to identify people’s core demands vis-à-vis the apartheid government. In terms of the content of the final constitution, however, direct public consultation in the constitution-making process had little impact on the substance, and there was no referendum to ratify the document. Instead, participation seemed to have affected the buy-in to the ultimate document and has bought South Africa’s leaders two decades of support for the constitution as a symbol, even if that support is fraying at the moment (IDEA 2016). Having said this, public involvement made itself felt in other ways, without need for invitation by the Constituent Assembly. As Ebrahim recounts, the role of the trade unions was significant in advocating for their interests with regards to the “lockout” clause, and this public pressure resulted in breaking a deadlock between negotiators.
Cosmetic participation was much more common in our case studies. In Kosovo, by the time the public process was launched upon the Declaration of Independence, much of the text was already worked out in draft. Furthermore, the need to move quickly meant that there was no real opportunity for input. Final approval by the representative of the international community was important but also meant that there was not even a formal process in which the public was able to bless the document.
In Kenya too, the process had formal room for public input, but this had little impact on the final text. The Committee of Experts did engage civil society, holding meetings and soliciting public input. But the prior round of constitution making had included a broader program of participation that informed the 2009–2010 process. Public debate mainly occurred after the draft text had been offered up for adoption, in preparation for an up or down vote in a referendum.
In Nepal, the second Constituent Assembly had a public outreach committee. But public consultations commenced only after the draft was published in 2015 and were limited to the capital district. No referendum was required or held. Since promulgation, the public in certain areas of the Madhes plains have made their voice felt through protests and a blockade of trucks arriving from India. This public pressure on the economy produced the first amendment to the constitution in 2016, but whether it will meet the long-term demands of the Madhesi protestors is uncertain at the time of writing.
In Burundi, a ratification referendum was held at the end of an elite driven process, and the campaign for the referendum lasted a mere two weeks. While it may have allowed the parties and the media to superficially inform the public of the contents of the constitution, it seems to exemplify Eisenstadt et al.’s point about the disutility of referenda as the sole mode of participation.
In Tunisia, the Constituent Assembly welcomed civil society organizations during one dedicated “open-house” day. Individual drafters also participated in public consultations throughout the country and among the diaspora, but the organization was so uneven that the initiative was more symbolic than substantive. The more influential form of public participation came in the form of spontaneous comments and posts on social media platforms. Tunisians managed to continue using social media tools beyond the revolutionary movement in order to weigh in public debates over the constitution. Correspondingly, politicians had an incentive to communicate though social media since the tool enabled them to access audiences and therefore bypass the traditional media sympathetic to the old regime.
V. Lessons from the Cases: The Role of Courts
In several of our case studies, courts played a role in policing the process and keeping it on track. South Africa was a forerunner in this regard, introducing the novel idea of a new constitutional court serving as a downstream constraint through its role as guarantor of the principles set out in the first interim constitution. This court initially denied certification, forcing some changes in the final version of the constitution as adopted. Similarly, in Nepal, the Supreme Court played a role in providing the final blow marking the failure of the first Constituent Assembly, which Dev calls a “natural death.” Whether natural or not, the process needed someone to declare the end of life, and the Court played this role. Kenya’s process design allowed a special court to resolve procedural disputes, but this was not needed in the end.
The availability of courts to serve as monitors in this way obviously varies from society to society. In some political transitions, the courts are seen as partisan or instruments of a prior regime; indeed, for this reason the South Africans created the Constitutional Court as opposed to letting the ordinary judiciary play the role of monitoring. In those rare environments in which the rule of law is established and operational, we can conceive of independent courts as an asset available to constitution makers. In other instances, the courts may be seen as partisan and so are appropriate targets of inclusion-oriented institutional reforms. In Georgia, for example, the corruption of the courts and lack of judicial independence were seen as major challenges to be addressed through constitutional reform. And courts can sometimes serve as the means by which constitutions are reduced to being on mere “life support” (Elkins et al. Reference Elkins and Kahler2009). Kiganahe’s chapter describes how Burundi’s Constitutional Court blessed President Nkurunziza’s undermining of term limits in 2015, perhaps the death blow to hopes of inclusive governance.
Though not elaborated in our case study, the Burundi case includes an incident subsequent to promulgation of the constitution in which twenty-two MPs who had left the ruling CNDD-FDD party to serve as independents were declared to have lost their seats by the Constitutional Court. This incident in 2008 provoked a political crisis and much criticism. The judicial role here was not so much to serve as an independent monitor but rather as a symbolic resource to advance a particular position, consolidating the power of the ruling party.
VI. Lessons from the Cases: Substance
A. Issues of Transition
A1. Transitional Provisions
If a constitution can be compared to the rules of a board game, transitional provisions are akin to rules regarding the initial organization of the game – they provide instructions on how to set up the various parts and pieces so that it is ready to play (Bisarya Reference Bisarya, Ginsburg and Huq2016). They usually include rules regarding the first elections, the establishment of different constitutional institutions such as a Constitutional Court and electoral commission, and rules regarding the status of existing legislation. Transition puts a high premium on controlling the first government and legislature, which will dictate the establishment of the constitutional infrastructure.
The example of Burundi is interesting in this respect. Burundi’s constitution treated the initial five-year period of implementation of the constitution as a special period. The first presidential election would be indirect, being carried out in the National Assembly, on the theory that the war might prevent a nationwide election (and also reflecting the interest of former president Buyoya in avoiding a national campaign). As noted above, this was an object of grave miscalculation by the G-10 and Buyoya. Pierre Nkurunziza, leader of the CNDD-FDD, was elected into office and has entrenched himself in power ever since.
Another Burundi innovation was to have special requirements for supplemental seats if one party earned more than a threshold. This served as a penalty on overly strong parties, on the theory that it was necessary to cement consociationalism in the first elections.
In South Africa, the interim constitution created a power-sharing government with the ANC, NP, and Inkatha Freedom Party all sharing in executive power until the first elections under the new constitution in 1999. Although the NP decided to withdraw in 1996, the idea behind the initial power sharing was to ensure that no party could have unilateral control over the early stages of constitutional implementation, while giving the ANC some political cover so that they were not solely responsible for the many difficult decisions the large-scale constitutional reforms necessitated, and also giving the NP some political leverage (although it turned out to be less than they had expected)
In Tunisia, the transition occurred with a couple of hiccups. The sequence of presidential and legislative elections was left out of the transitional provisions, leaving this decision to the political bargaining process. Based on Article 89 and the spirit of the constitution, since the elected president is required to ask the winning majority in Parliament to name a prime minister to form a government, it makes sense that the presidential elections take place first or at the same time of the legislative elections. However, since Ennahdha and Nidaa Tounes parties both saw an advantage if legislative election were held first, the presidential election took place later, which further delayed the formation of a government.
Kosovo, on the other hand, was forced to complete its transitional period under international supervision of its nascent political institutions. It is interesting to note from the case study that once period this ended, in 2012, the Kosovars amended the constitution to delete the transitional provisions from the text – removing the trace of diminished sovereignty in the formal document.
A2. Implementation
Major constitutional dramas like the cases discussed here often involve a good deal of institutional reform. This leads to the issue of how constitution making is balanced with the need for institutional and legal reform at the subconstitutional level, a particularly important issue during political transitions. Kosovo’s constitution-making process had to be completed within a 120-day period, as mandated by the Vienna talks, but much other legislation had to be adopted in this period as well.
The Tunisian case was perhaps emblematic of what happens when a country emerges from a dictatorship with extremely concentrated powers. Drafters in such contexts may believe that multiple institutions can serve to strengthen checks and balances, and Tunisia created five such constitutional bodies in addition to the Constitutional Court and the Supreme Judicial Council. As we write, almost ten years after the adoption of the constitution, only two of these bodies had actually been created, and the Constitutional Court has not been formed because of political obstruction. In this case it would have been more effective if the Constituent Assembly had created all these bodies before the end of its mandate in order to ensure that the transition does not suffer a breakdown.
Kenya’s process included a novel institution, the Commission for the Implementation of the Constitution, tasked with overseeing the transition and the establishment of new institutions. The Commission was to have a lifespan of five years, and reported to a parliamentary committee, but also issued numerous public reports. This novel idea had first surfaced in a proposal by Yash Ghai in his work on the Constitution of Afghanistan, but the body never really played its expected role in that country. In Kenya, however, the Commission did play a role and ensured some accountability.
B. Preambles and Other Symbolic Issues
Preambles are the “mission statements” of the constitution and can be powerful symbols of the constitutional order and identity (King Reference King, Galligan and Versteeg2013). They explain why it is that constitution making is taking place and what its goals are. They often go through long historical sections, listing the grievances and glories of the nation. While not always legally enforceable, they raise great symbolic issues that can often be among the most contentious. Yet to date, we have few accounts of how these sections of constitutions are actually drafted.
Our cases present a range of experiences in terms of how the preamble is produced. In the case of Kosovo, the definition of the nation itself was highly contested, since the predominately Albanian country contained a Serb minority that was not participating in the process. Recounting the contested and emotional history required a good deal of effort, to avoid triggering a backlash from either side of the cleavage. However, when the draft leaked, the international community grew concerned, and pushed for to the postponement of the issue. The need for approval by the International Civil Representative provided limits on what could be said, and in particular led to the dropping of historical references. The final version was watered down, providing an example of what Caka calls the “passion management” of the international community. At the same time, the preamble signals its international orientation by referring to the process of Euro-Atlantic integration. (This phrase had originated as a title in Romania 1991, was then used in preamble of Montenegro 2007, followed by Kosovo, and later was proposed for Georgia’s constitution.)
The preamble of the Burundi Constitution was designed to foreshadow the themes of the constitution, while also invoking the recent history. This made sense to do at the end of the process, which we learned was a typical approach in these cases. Nepal’s preamble was mainly drafted in the second Constituent Assembly, and generated extensive debate over concepts like proportionate inclusivity, pluralism, and patriarchy. The process involved active debate over values.
In South Africa, the preamble was seen as a potentially contentious topic, put into the “parking lot” so as not to derail the negotiation of more functional provisions of the constitution. Each of the eleven sequential draft texts lacked a preamble. When it came time in the last three months to draft the preamble, disagreements around the treatment of God threatened to create a deadlock. The solution was to form a two-person drafting committee, composed of the pastor for the National Party and the Secretary of Education for the Communist Party. These ideological opposites were able to craft an eloquent and accessible preamble, embodying the overall spirit of compromise which then prevailed.
Tunisia’s preamble was heavily debated, particularly over the balance between religious and secular references. Aware of South Africa’s approach of leaving the preamble to the end, the drafters decided to give it to the relevant constitutional committee rather than leave it to the coordinating committee at the end or deciding it in plenary. And in Ecuador, President Correa insisted on including the name of God in the preamble, but the preamble’s text was excluded from that put before the public in the approval referendum.
Preambles are not the only symbolic issues that can become controversial. Mottos, flags, and national anthems also are imbued with meanings that outsiders may not fully appreciate. Interestingly, in both Nepal and Kosovo, the national anthem was produced through an open competition. In Kosovo, the issue was so riddled with political sensitivities that the national anthem lacks words. The design of the national flag too was contentious – most Kosovars identified with the colors red and black, those of the flag of Albania. But sending such an exclusionary message regarding the Kosovar Serbs was out of the question, and so blue and yellow, the colors of the European Union, were chosen instead, much to the surprise of the general population. Of note too is the flag design, which called for stars that would reflect the number of communities in the country; however, because the majority Albanian community is not mentioned, it was left out of the flag, and thus the seven ethnic groups have six stars, with the largest group being unrepresented!
The bottom line is that symbolic issues can be extremely important in helping to bring people together. Compromise approaches, such as that of the Kosovo flag or Nepalese anthem, may work to avoid blowing up the entire constitution-making process. But they probably do not help achieve a sense of buy-in among the public. Instead, they present examples of pitfalls that were successfully avoided.
VII. Lessons from the Cases: Innovation and Borrowing
Another theme in the recent literature to which our cases respond is the role of innovation and borrowing. Constitution making is a process in which there is a good deal of diffusion (Law and Versteeg Reference Law and Versteeg2011; Elkins Reference Elkins and Kahler2009) in terms of content. Notwithstanding a normative view that constitution making should be a local process, institutional designs and ideas flow freely across borders.
South Africa, of course, is a central case of innovation on both process and substance (Arato Reference Arato2016) and has led to a number of core ideas in the lexicon of constitution making. In terms of substance, it introduced a unique form of indirectly elected executive president; a complex scheme of subnational governance that was nevertheless not federal; and instructions to the judiciary to interpret the bill of rights in light of international and comparative practice. Many of these innovations were adopted after extensive study of other cases, including India, Canada, and Germany with regard to the Upper House. But some were truly worked out on the spot, including the idea of a postamble in the interim constitution. Some, though not all, of these innovations have diffused to other countries, and the prominence of several South Africans (including our authors Ebrahim and Murray) in the field of global constitutional advice is perhaps both a reflection of its importance and a mechanism of diffusion.
Burundi’s constitution included a novel design of what we might call modified consociationalism (Reyntjens Reference Reyntjens and Kuperman2015). It mandated that no more than 60 percent of the Parliament could be from one group, effectively capping Hutu representation at that level, and instituted a two-thirds majority for legislation so that each major group would have a veto on policy. It also required two vice presidents, one from each ethnic group; equal representation for the groups in the Senate and armed forces; a cap on the mayoralties held by one tribe; and other features. But it also had some integrative requirements, including the idea that parties had to be multiethnic. Out of every three members, one had to be from the other group. This combination of institutions derived from some prior Burundi constitutions, but also reflected creative proposals by the mediators, leading to a distinctive combination of institutions. (While not the focus of our analysis, it is worth pointing out in passing that the eventual unwinding of the Burundi bargain had little to do with the substance of its consociational solution. Rather, the parties that were called on to implement the constitutional bargain were not those that had negotiated it.)
Nepal’s environment is one in which there is a good deal of openness to foreign ideas, but also a distinct emphasis on localization. The concept of “proportionate inclusive democracy” was seen as being a distinct one in South Asia, though of course it borrows elements commonly found in the region. Sometimes innovation occurs through bricolage (Tushnet Reference Tushnet1999). Perhaps the most significant borrowing in Nepal was the idea of federalism, which had been introduced into debate as early as the 1950s. While there is no doubt that federalism has been a persistent demand of some parts of the political spectrum in recent years, it is not clear that any federal scheme could live up to the demands being put on this one. The fact that borders and names must be produced from scratch only adds to the challenge (Stepan Reference Stepan1999; Choudhry & Anderson Reference Choudhry and Anderson2019).
Tunisia’s most famous innovation was to allocate eighteen seats in Parliament to the 10 percent of its population living abroad. The representation of the diaspora was supposed to be an exceptional measure for the Constituent Assembly only, since the constitution could not exclude such a large group of citizens. However, the measure was kept in the final constitution. Inspired by Tunisia, France adopted the measure in 2012. Less known novelties include the right to privacy, which was a way to protect citizens from mass surveillance; the obligation for the state to protect the climate; and the effort to reduce inequality by applying a principle of positive discrimination between regions. A more controversial idea was the mention of Palestine; it is rare to name a country other than a former colonizer or kingdom in a constitution.
Kenya’s innovations were both procedural and substantive. The technical Committee of Experts included three foreigners, including our author Christina Murray, which is a relatively rare, but in this case effective, way of proceeding. Substantively, the constitution’s elaboration of principles, introduction of judicial vetting, and commitment to good governance marked a significant break with Kenya’s prior constitutions, as well as synthesis of various ideas present in the field in the early twenty-first century.
VIII. Conclusion: Updating the Elster Framework
Working through the details of these cases provides some thoughts on how well Elster’s framework of Forces and Mechanisms maintains its relevance in modern constitution making. The central notions of the Elster framework remain compelling – firstly, that constitution making is a bargaining process, and secondly that constitutional design choices are not completely free, but are shaped by constraints which exhibit their effect, before, during, and after the constitutional negotiations. His identification of reasons, passions, and interests as powerful forces in constitution making holds up well, even if one can quibble with the optimal balance among the three (Brown Reference Brown2008).
However, it is worth considering how the balance among different forces may have changed in recent waves, and whether recent experience requires some updating of the framework. Several elements bear analysis, including the notion of constraints, the role of external actors (specifically the public and the international community), the effect of political parties and elections, and the iterative nature of most constitution-making processes.
A. Constraints
Constraints on decisions regarding constitution design choices remains a useful concept, but we propose a slightly different understanding of constraints than the one proposed by Elster. Firstly, Elster states that “constraints that do not constrain are not constraints” and cites the example of the Philadelphia Convention, which was convened in order to amend the Articles of Confederation yet proceeded to ignore this constraint by writing a new constitution. While the constraint may not have been effective, it certainly affected decision making and argument, and ignoring it may dampen our understanding of the overall forces and mechanisms at play. No constraint is ever failsafe; rather, it is more accurate to describe constraints as heightening the political (or other) costs of certain actions or decisions. In Kosovo, for example, the certification of the ICR was clearly a downstream constraint on the choices of the constitution makers, just as the Ahtisaari principles were upstream constraints, notwithstanding the fact that in hindsight we know that drafters chose to ignore one particular principle about suppressing mention of Albanian identity.
Secondly, there are a range of general contextual influences which significantly shape the constitutional bargain. These are not true constraints but frames and models. Issues such as preferred constitutional design choices in the geographic region, colonial history, countries where legal elites are educated, and the specific drama(s) which gave rise to constitution making will all shape the preferred options of the constitutional negotiators. These contextual considerations can affect constitutional decision making in at least a couple ways. For one, they can frame the possible universe of options for constitutional design choices within which negotiations can take place. For example, in Tunisia the influence of France’s system of semi-presidentialism – which had been the previous form of government and was a system that political actors and lawyers understood well as many of them had studied in France – was a considerable influence on the discourse concerning system of government in the NCA. In addition, appeals to contextual factors such as history may be used by one side to strengthen its argument during bargaining. The phenomenon of “restoration constitution making” provides an illustration of this, whereby appeals to a democratic past led to the restoration of pre-Soviet constitutions in the Baltic states, which was used by nationalistic elites to deprive Russian-speaking populations of citizenship rights (Partlett 2015). Aversive constitutionalism, in which anti-models motivate choices to avoid, is another (Scheppele Reference Scheppele2003). Contexts can turn frames and models into resources. These influences cannot be ignored if we are to describe fully the forces and mechanisms at play in constitution making.
Thirdly, and related to the last point, the term “assets” is a useful umbrella term to describe preexisting factors which facilitate completion of the constitutional bargain. Each party has its own set of assets (and constraints) that may be used to favor one constitutional design choice over another, but we are interested in those which shape the process as a whole. If it is the process that is the unit of analysis in modern literature (Miller and AuCoin Reference Miller and AuCoin2010) then it may be helpful to think systematically across cases about the factors that facilitate and retard constitution making. This idea is consistent with Elster’s general point that the constitution-making process is socially situated, but goes further to note that the environment does not merely constrain but can provide powerful resources. A systematic analysis of individual cases should identify constraints but also assets.
B. External Actors
B1. The Public
The public appear in the Elster framework only as a downstream constraint through approval or rejection of a constitutional draft at referendum. However, the role of the public in constitution making has expanded significantly over the past twenty years and is worth examining in more detail. In addition to serving as a potential ratifier of a constitution, the public plays a number of other roles in constitution making, all of which can be considered in terms of the forces and mechanisms exerted on decision making. The public can serve as an initiator of constitution-making episodes, as a selector of the drafting body, as advisor and source of ideas, and – in some rare modern cases – as a direct author. The public can also be a force to be manipulated, as the emerging practice of the instrumentalization of public support in constitution making by populist authoritarians (Landau Reference Landau2013; Partlett Reference Partlett2012).
With regards to the public as initiator, we refer to the particular case of popular revolutions whereby an uprising consisting of a significant number of citizens results in the overthrow of the previous regime, and among the demands of the revolution is constitutional change. In this case, the public can act as an upstream constraint on constitution making – it assumes the constituent power and convokes the constitution-making body, and often places certain demands on the content of constitutional change to align with the overall goals of the revolution (Preuss Reference Preuss2008). A recent example is the EuroMaidan revolution in Ukraine which not only forced then-President Yanukovych from power but also demanded constitutional change in the form of a return to the constitution as it stood in 2006 (returning to a premier-presidential system of government) (Choudhry, Sedelius and Kyrychenko Reference Choudhry, Sedelius and Kyrychenko2018).
There are other examples of the public (in the sense of nonpolitical actors) acting as an upstream constraint in the absence of revolution. Preuss (Reference Preuss2008) describes how the industrial leaders and trade unions in Germany negotiated the “Stinnes-Legien Pact,” which recognized unions and the right to strike in exchange for the abandonment of a push for a socialist economy. Given the power of both groups to thwart the work of the Constituent Assembly, this pact “predetermined essential parts of the content of the future constitution.” The role of the industrialists and union leaders in the making of the Weimar Constitution is thus aptly described as “pre-constitutional guarantor” (Preuss Reference Preuss2008: 115).
The public’s role as selector refers to the elections of a constitution-making body. As envisioned by Elster, the convener of the constitution-making body acts as an upstream constraint, but the mechanics of elections – as opposed to convocation by a king, as in the example of the revolutionary French Constituent Assembly described by Elster – nuance the way in which the upstream constraint operates. This is explored further below in Section VII.C on Elections and Parties.
Perhaps the most significant evolution in the role of citizens in constitution making has been through the use of public consultations, which usually takes the form of general town hall meetings, written submissions, or consultations with specific interest groups. As constitution makers consult, but are not bound by, public opinion, we describe this role as advisory. We would also include in this category lobbying, advocacy, protest, and other forms of public expression aside from voting. There are two ways to look at the role of the public in terms of the Elsterian framework: One is that public participation can constrain constitution makers to negotiate. For example, in Tunisia four civil society organizations, primarily led by trade unions, convened a “national dialogue” to force the main parties to break the deadlock and come to a deal regarding the resignation of the government and finalization of the constitution. Looking outside of the cases we have examined here, in the process begun in Chile in 2015 the center-left oriented government sought to leverage an extensive public participation exercise to generate broad-based interest and support for constitutional reform, which might constrain reluctant parties on the right to come to the negotiating table.
Elements of the public can also play a key role in the arguing and bargaining over interests, which may be critical in times of deadlock. To mention two examples from the cases described here: In South Africa the deadlock over the “lockout” clause was resolved in favor of the ANC thanks, in great part, to pressure from mobilized trade unions. While in Tunisia, large street protests of women were mobilized in relation to the unequal standing of women described in the first public draft of the constitution. This encouraged the Islamist party Ennahda to cede on this point to secular/liberal parties at the negotiating table. In both of these examples, the arguing/bargaining process was influenced by actors external to the negotiating forum, resulting in one side being able to assert its preferred constitutional design choice.
In rare cases, the public might itself serve as author of the constitution. The interesting experiment in Iceland after the financial crisis, in which a constitutional draft was produced in a process that specifically excluded elected politicians, has inspired analysis and imitation (Gylfason Reference Gylfason, Ingimundarson, Urlfalino and Erlingsdóttir2016; Landemore 2013, 2016). Recently, Mexico City adopted a substate constitution that also relied heavily on ordinary citizens as drafters (Houlihan and Bisarya Reference Houlihan and Bisarya2021: 21).
Given the general confinement of the role of the public to ratification referendums, which are becoming more popular but may matter less in terms of channeling effective participation (Eisenstadt et al. Reference Eisenstadt, Carl LeVan and Maboudi2017), the Elster model might lead to us to conclude that the role of the public is not a significant factor in the forces and mechanisms of constitution making. But this is far from the truth. A closer examination of the role of the public adds a multidimensionality to the framework for constitution making not present in Elster’s original framework: The public can act at various stages of the process. Its role may vary from case to case – from an upstream constraint to a deadlock breaker to influencing the negotiating positions of constitution makers – but it is important to include in any updated model of constitution making.
B2. The International Community
The international community is an amorphous concept, but of significant influence in constitution making today. Based on the cases, we can identify several ways in which international influence comes to bear on national forces and mechanisms of constitution making.
As a preliminary point, it is important to note the variety of roles the international community can play in constitution making. Such roles include a dominant, primary role such as that of the United States in the development of the post–World War II constitutions of Germany and Japan or, like the ICR in Kosovo, having decision-making powers but allowing national actors to take the primary drafting role. In some contexts, such as Nepal and Tunisia, international advisors were a constant feature but had no decision-making powers. In other places, international actors have no formal decision-making powers, but do have highly powerful leverage (such as Iraq, Afghanistan, and very often in the work of the Venice Commission). There are also contexts in which international actors do not have any direct, active role but one can still discern constraints or assets driven from international influence (e.g., the need for international legitimacy in South Africa, or the pressure brought from international actors on Kenyan elites to bargain), as well as situations where international advisors are present to provide information to one or more sides.
C. Elections and Parties
While there is a good deal of conceptual and empirical work on how elections held after the constitution-making process can alter incentives and bargaining positions during constitutional negotiations (Ginsburg 2003; Negretto Reference Negretto2013), there is relatively little scholarship on elections to constitution-making bodies. Yet the selection of the constitution-making body forms a key part of the Elster framework, in which the selector, and the mandate given to the selected, act as upstream constraints on the process.
While not always the case, elections have become the most common means of selection for constitution-making bodies, and there is much more to explore in how elections can differentially affect upstream constraints. Closely related to this issue is the presence and functioning of political parties, which did not exist in the cases examined by Elster in his classic paper but can have a significant effect on the forces and mechanisms of modern constitutional politics.
In our cases, the constitution-making bodies in Nepal, South Africa, and Tunisia were elected as dual-purpose bodies – legislatures and constituent assemblies. In Kosovo and Georgia, the elected Parliaments only rubber-stamped the constitutions which were drafted by small commissions, whereas in Burundi no elected body took part in the drafting process.
Tunisia is illuminating with regards to the constraints caused by electoral systems as, by many accounts, there was consideration of two alternatives: the proportional system, which was eventually chosen versus a majoritarian system. Most commentators believed that the latter would have greatly favored Ennahda, perhaps giving it a majority in the Assembly and enabling it to drive the agenda. Instead, the proportional system resulted in an Assembly in which no party had a majority, and instead Ennahda formed a three-party government with two secularist parties – and consensus within this “troika” was then necessary to finalize the constitution. The constraints caused by this consensus requirement among several parties almost led to the collapse of the Assembly and the process, rescued only by an important intervention from civil society actors, the establishment of the Consensus Committee which included all parties on an equal basis, and a new government.
In South Africa, the elections of 1993 meant that the ANC would have a majority in the Constituent Assembly, but it fell short of the two-thirds needed to pass a constitution on its own; it was thereby constrained to find a negotiated consensus rather than push unilaterally for its own agenda.
In Nepal, the same electoral system resulted in two very different Constituent Assemblies. The first Constituent Assembly was highly inclusive, perhaps overly so with 25 parties represented, and bridging consensus over the required 401 members proved an impossible task. In the second Constituent Assembly, the Communist Party Nepal – Unified Marxist Leninist (CPN-UML) and Nepali Congress parties found themselves very much in control, winning 388 seats and requiring only a few more small parties to reach a constitution-making majority.
D. Iterative Constitution Making
Arato (2009) laments the failure of the Iraq constitution-building process for fear that it would lead subsequent countries to move away from a potentially influential model for modern constitution making. The model he was referring to consisted of two main stages: a closed, elite-driven roundtable process which agreed upon an interim constitution, which would then bind a freely elected Constituent Assembly that would draft a “final” constitution. Arato need not have worried, as phased constitution making has become a common occurrence, and one that is celebrated in certain contexts (Lerner Reference Lerner2011). In the cases that follow, South Africa and Nepal feature this model. Kenya offers a different form of iterative constitution making in which an earlier process resulted in a draft which was rejected at referendum, but many of its findings and ideas were picked up by a subsequent process.
This greatly complicates our framework of constitution making. Under Elster’s model, there was one, principal site of constitutional negotiations, subject to constraints based on decisions taken before its formation, and decisions to be taken after completion of its work. During its work, there was one process of arguing and bargaining, with players driven by passion, interest, and reason, towards one set of decisions accumulated into a draft constitutional text.
Modern constitution making, in particular in post-conflict contexts, often consists of several rounds of decision making, each building upon those before it, and preparing the ground for those to come subsequently. This creates a complex dynamic of constraints, some acting in chain-like fashion from one stage to the next, and others remaining a constant pressure on incentives throughout. While iteration can produce mutual trust among negotiators over time, it also extends constitution making temporally, rendering it vulnerable to shocks and surprises that might upend prior rounds of agreement. It is telling that in so many of our cases, the issues that ended up creating barriers were not those anticipated in the outset.
Forces and mechanisms; assets and constraints; reason, passion, and interests. In an era of more transnational involvement in constitution making, the set of players and forces is more diverse, the temporal frame more extended, and the balance among factors more contingent and complex. What is enduring is the ongoing search for stable constitutional order, and the need to understand how the participants view the process from the inside. The remainder of this book pursues this task for several recent cases.
