I. Introducing Constituent Assemblies

The contributors to this volume examine constituent assemblies in a number of countries and historical periods, from different comparative and theoretical perspectives. In doing so, they join a wave of studies of constitution making that has gathered momentum over the last decades (see Choudhry and Ginsburg Reference Choudry and Ginsburg2016; Miller and Aucoin Reference Miller and Aucoin2010; and Voigt Reference Voigt2013 for some previous edited volumes). The present book supplements these studies in several respects. In this Introduction, we discuss some salient themes and place them in context; we also refer to constitutional episodes not discussed in this volume.

The chapters have a large geographical breadth and historical depth. Going beyond the much discussed (and still fertile) questions posed by constitution making in Europe and North America, five of the nine chapters in the volume discuss constituent assemblies in Latin America, Africa, and Asia. Concerning Latin America, the chapters by Gargarella and Negretto range from the early nineteenth to the early twenty-first century, and cover constitution making in Argentina, Bolivia, Colombia, Chile, Ecuador, and Venezuela. Concerning North Africa and West Asia, Lerner’s chapter includes case studies of constitution making in Egypt, Tunisia, and Israel. Concerning South and Southeast Asia, the chapters by Bhatia and Malagodi address constitution making in India, Pakistan, and Nepal, while Lerner, in her chapter, includes case studies of Sri Lanka and Indonesia, as well as of India. Elster’s chapter on the making of the Norwegian 1814 constitution and Gylfason’s chapter on the process of constitutional reform in Iceland that began in 2010 offer case studies of these two small Nordic countries. In their chapter, Bucur and coauthors cite France, Ireland, and Romania as the main cases illustrating their general thesis. At the subnational level, the American state constitutional conventions that Negretto discusses in his chapter constitute an invaluable resource for comparative analysis.

The word “assembly” needs some context. First, and most obviously, a constitution made by an assembly differs from that handed down by a single individual. In Plutarch’s Parallel Lives we find stories about how Solon and (more uncertainly) Lycurgus wrote the constitutions for Athens and Sparta respectively. Descartes (1637, § 2) argued that this procedure has the advantage of enhancing the coherence of the document: “If Sparta was in earlier times very prosperous, that was not on account of the goodness of each of its laws in particular, seeing that several were very strange and even contrary to good morals, but on account of the fact that they were devised by only a single man and thus they contributed towards the same end” (italics added). In recent times, the 1958 constitution of the Fifth French Republic comes closest to this model. In his speech in Bayeux in 1946, de Gaulle had already laid out all the basic principles of the 1958 constitution, except for the procedure of referendum (Maus Reference Maus, Maus, Favoreu and Parodi1992, 262). All institutions were to serve the overarching goal of creating a strong executive. While de Gaulle did work with a constitutional assembly in preparing the document, its role was merely consultative. Although the upstream instructions from the parliament of the Fourth Republic prevented him from adopting an American-style presidency, this was not his preference in any case (Peyrefitte Reference Peyrefitte1994, 446). There is no indication that the downstream constraint on ratification by referendum interfered with his freedom of choice, in the way ratification by state electoral assemblies constrained the choices open to the Federal Convention (Amar Reference Amar2005, 279–80).

Second, we need to distinguish genuine constituent assemblies from sham assemblies. Consider the following statement, sometimes imputed to Napoleon: “Il faut qu’une constitution soit courte et obscure. Elle doit être faite de manière à ne pas gêner l’action du gouvernement.” (“A constitution should be short and obscure. It should be written so as not to interfere with the action of the government.”) It is understandable that an autocrat would want to be unhampered by rules and restrictions, but the criterion for a constitution being genuine is that it does impose limits on the government; otherwise it is a mere pretense. In other words, either the upstream actors or some of the framers must be motivated to write a genuine constitution for the assembly to be considered genuine.Footnote ¹ Hence the bodies that adopted Soviet-style communist constitutions, for instance, should not count as constituent assemblies.

II. Internal Rules and Procedures

Except for the 1958 French case, all modern (nonsham) constitutions have been made by elected assemblies. Often, however, they have been prepared by unelected bodies or informal groupings. In her chapter, Lerner argues that the study of the constitution-making process in divided societies “should pay close attention to the politics that preceded the formal stage of drafting,” and shows that in her six case studies the success or failure of the formal process correlated highly with the presence or absence of informal talks and negotiations. Beyond her case studies, the 1989 Round Table Talks in Poland and Hungary also shaped constitution making in crucial ways, and may have been necessary conditions for the peaceful transitions that took place (Elster Reference Elster1996). These cases should be distinguished from those in which informal bargaining substitutes for, rather than prepares the ground for, the formal process. In her chapter, Malagodi discusses the making of the 2015 Nepal constitution as a case in which the assembly was largely bypassed.

In modern constitution making, it can be misleading to assert that constitutions are made by assemblies, if by that expression one intends to say that they emerge solely and organically from plenary debates among independent delegates. Although this characterization partly fits the two Nordic assemblies discussed in the chapters by Elster and Gylfason, it is less adequate for the other cases in this volume. For one thing, many crucial decisions are elaborated in committees rather than in the full assembly. For another, delegates are often not independent, but subject to party discipline. Since committees as well as political parties tend to be black boxes, in the sense that we usually know little about their internal decision-making processes (see Martin and Rasch Reference Martin, Rasch, Müller and Narud2013 for the opacity of parties in constitutional change), our understanding of the causal chains that lead up to the adoption of the final document may be incomplete.

The size and duration of constituent assemblies vary, and can matter. The smallest constituent assembly on record is the Icelandic one, which Gylfason describes in his chapter. Its twenty-five members could engage in even more focused deliberation than the fifty-five members of the Federal Convention (not all of whom were present at all times). The debates among the 112 members of the Norwegian assembly that Elster considers in his chapter on 1814 also seem to have been orderly. At the other end of the spectrum, the 1789 French assembly counted around 1,200 members. As most debates took place in plenum, chaos reigned. Gouverneur Morris (Reference Morris1939: 382), one of the most active members at the Federal Convention who was in Paris during the sitting of the Constituante, describes the proceedings as follows:

Another foreign observer, the British agronomist Arthur Young (Reference Young1794, p. 283), was equally shocked:

In the large modern assemblies discussed in this volume, chaos was probably avoided by delegating much of the work to committees.

The size and duration of constituent assemblies are, at least roughly, correlated. The small Icelandic assembly debated for four months, the Constituante for more than two years. Since the size of the assembly is also roughly correlated with the size of the country, the latter variable may in fact be decisive: the Indian assembly, with fluctuating membership between 200 and 300 members, took more than three years to complete its task. A large country may have to address more issues than a smaller one, and require a larger number of delegates to ensure knowledge about local conditions. Other variables, too, can affect the duration of the process. In particular, as Elster argues in his chapter on 1814, a tense international situation may lead to a speeding up of the process. In France in 1958, the urgent need to solve the Algerian problem had the same effect. More mundanely, at the Federal Convention many delegates were in a hurry to get home to their families and businesses. In India, external and internal threats had the opposite effect – delaying the assembly’s progress by more than two years past a self-imposed deadline.

In a tradition initiated by Sieyes (Reference Sieyes, Lembcke and Weber2014, 126–7), the hallmark of a constituent assembly is often said to lie in its unrestricted power. A constituent power cannot be bound (he argued), since the organ binding it would have to be one of the constituted powers that the constitution is to regulate. To allow it do so would be to have it act as judge in its own case. If, for instance, the organ convoking a constituent assembly tries to impose procedural or substantive constraints on its work, it is free to ignore them. A famous example is the decision by the Federal Convention in Philadelphia to ignore the instructions from the Continental Congress with regard both to substance (the Convention took it upon itself to create a wholly new constitution rather than simply revise the Articles of Confederation) and to process (the proposed document would be approved by specially elected ratifying assemblies rather than, as required in the Articles, by the state legislatures). The latter act of disobedience is especially important. Since the proposed constitution would have reduced the power of the state legislatures, they would hardly have accepted it had they been allowed to act as judges in their own case. As Negretto explains in his chapter, a similar refusal to respect upstream constraints occurred in Colombia in 1990. In that case, however, it was the Supreme Court rather than the assembly itself that declared that the framers would be free to decide on the content of a new constitution. In her chapter, Malagodi shows that Nepal’s constituent assembly ignored upstream instructions that imposed a two-year deadline on its work by extending its term four times.

Some cases contrast markedly with the sovereign assembly model. As noted, in 1958 the French parliament did not give de Gaulle a totally free hand to propose a new constitution. Several chapters in this volume also show the limitations of the alleged sovereignty of the constituent power. Gylfason’s analysis of the obstructionist role of the Icelandic parliament offers the most striking example. Since a major aim of the committee that drafted the new constitution was to eliminate the overrepresentation of the rural districts in parliament, the majority of that body had an incentive to block the reform and did so. In his chapter, Negretto shows that the making of the 1994 Argentine constitution was substantively constrained by a congressional law based on a political pact between the government and the opposition. As Elster notes in his chapter on Norway, the framers respected procedural instructions laid down by the prince-regent.

A recurrent issue in constitution making is the double role of electoral systems. On the one hand, delegates are usually chosen by popular elections, the main exception arising in federal systems in which state legislatures may select the delegates. This was the case for the Federal Convention, for the German convention that adopted the 1949 constitution, and, as Bhatia explains in his chapter, for the 1946 elections to the Indian assembly. On the other hand, one of the central tasks of the assembly is to design or at least sketch in broad outline an electoral system for future legislatures. Although not all constitutions specify the system in great detail – the 1958 French constitution is very laconic on the subject – many do. Tensions may then arise if the franchise used in electing the framers differs from the franchise they write into the constitution. In France in 1789, Robespierre (Reference Robespierre1912–67, Vol. VII, 172) very effectively criticized the assembly for adopting stricter tax qualifications for suffrage than those under which they had been elected, arguing that in doing so they retroactively undermined their own legitimacy. As Elster notes in his chapter on 1814, a similar discrepancy arose in the Norwegian case, although nobody (to our knowledge) commented on it at the time. In his chapter, Bhatia notes a converse kind of tension between the qualified suffrage used in electing the provincial legislatures that nominated the delegates to the Indian assembly and the universal suffrage adopted in the constitution. How, some asked, could an elite body create a democratic constitution?

In his chapter, Bhatia also cites a somewhat similar puzzle, which to our knowledge has no parallel elsewhere: how could a unicameral assembly legitimately adopt a bicameral system in the constitution? If the delegates adopted bicameralism because of its superior cognitive features, would they not, as Robespierre argued with respect to the suffrage, retroactively undermine their own legitimacy? There are examples, however, of bicameral constituent assemblies: Japan 1946, Turkey 1961, Sweden 1974, Spain 1978, Romania 1991, and Poland 1992. With the exception of Sweden, none of these adopted a unicameral constitution.

In many countries, constitutional amendments require, for their adoption, a qualified majority. It might seem paradoxical, therefore, that the more fundamental decision to adopt the constitution that is being amended is virtually always done by simple majority voting. There is no paradox, however, if there is no constitution in place that can serve as a default option if a required qualified majority is not reached. (For a similar reason, parliaments always adopt budgets by a simple, and in some cases absolute, majority.) Conversely, the requirement, noted by Negretto in his chapter, that the 2007 Bolivian constitution be passed by a two-thirds majority in the assembly should not surprise us. This being said, we may well wonder why in other cases a simple majority is used even when a default constitution is in place. The answer may be found in the degree of malfunctioning of that default system.

The use of simple majority voting is consistent with a desire for a large consensus, to enhance the legitimacy of the new constitution. In 1949, for instance, the German framers set themselves – and achieved – the aim of adopting the constitution by an 80 percent majority. The nonconsensual adoption of a constitution can tear up the social fabric, as shown by the adoption of the violently anticlerical 1931 Spanish constitution by a 53 percent majority. In her chapter, Malagodi discusses how the constituent process in Nepal involved a baffling mix of simple majority voting, the requirement of two-thirds majority, and a demand for consensus.

III. Social and Institutional Context

When do the people speak in the constituent process? According to the canonical formulation by Sieyes, the constituent power ultimately belongs to the Nation (as he said) or to the people (as we say). If the constituent assembly elaborates the constitution, how can the people at large shape it?

First, the people can speak at the beginning of the process, by choosing the delegates to the constituent assembly. In modern times, the choice is usually based on their programs or known opinions, but in some earlier assemblies (France 1789, Norway 1814) the deputies were elected on the basis of their character. As just noted, in India only some of the people spoke, and then only indirectly, through the provincial legislatures. In Norway, too, only some of the people spoke, but in direct elections. Today, universal suffrage ensures that the people as a whole talk when electing delegates to a constitutional convention or to a legislature that, as in France in 1946, is mandated also to propose a constitution. In his chapter, Negretto cites several other examples of such mandated assemblies, making the additional point that some may also be constitutionally authorized.

Speech at the beginning of the process can also take the form of imperative mandates from voters to their representatives. In constitution-making processes, this speech mode seems to be rare. As Elster discusses in his chapter on 1814, some of the resolutions of the Norwegian electoral assemblies had the flavor of mandates, but they were not binding. In 1787, the delegates to the Federal Convention from the small state of Delaware came with instructions not to assent to “the abolition of the fifth article of the confederation, by which it is declared that each state shall have one vote” (Farrand Reference Farrand1966, Vol. I, 6). The delegates followed the instruction, which was arguably decisive in forcing the large states to accept an equal number of senators for all states. It issued, however, from the state legislature, not from the voters. The grievance books that were written at the time of the elections to the Estates-General in France in 1789 contained many imperative mandates, which Louis XVI disallowed. They delayed the proceedings in the early stages, but probably did not shape the decisions (Hyslop Reference Hyslop1968, 99–104).

Second, the people may speak at the end of the process, by ratifying assemblies (as in America in 1787) or in a referendum (as in many of the other cases discussed in this volume). In a nonnegligible number of cases, the proposed document was rejected by the people (see Elster Reference Elster2013, 234 for some examples). In their chapter, Bucur et al. discuss what is arguably the most important case, the rejection of the first draft of the 1946 French constitution. One reason why it was rejected was probably the perceived importance of institutional interest behind the proposal: the constituent legislature created a “régime d’assemblée” that the voters found excessively legislature-centric. Yet even when the people speak approvingly at the end, its voice is not necessarily heard. When the referendum is purely consultative, the parliament may choose to ignore it, as happened in Iceland. As Gylfason recounts in his chapter, when Jon Elster visited Iceland in 2012 he confidently predicted that “If the people approved the constitutional proposal, Parliament would find it difficult to override the moral authority of the people.” In making this claim, Elster was influenced by the fact that the Norwegian parliament had respected the negative outcome of two consultative referendums on Norwegian membership in the European Union, in 1972 and in 1994, despite the constitutionally required majority in the parliament in favor of entry.

Third, there is a recent tendency for the people to be allowed to speak during the process, by submitting proposals to the constituent body and even, as Gylfason explains in the case of Iceland, to engage with the framers through an interactive website. If, as seems possible, this procedure becomes the norm, it would contrast with the claim by Elster (Reference Elster2013, 234–6) that an optimal constitution-making process should be “hour-glass shaped,” with upstream and downstream popular voices separated by an assembly deliberating behind closed doors. The proper mix of secrecy and publicity in debates and votes in the constitution-making process is a complex issue (Elster and Le Pillouer Reference Elster, Le Pillouer and Elster2015).

Several contributors focus on the difficulties of constitution making in, “plural societies,” that is, societies that are “deeply divided” on national, regional, ethnic, religious, or linguistic grounds. In his chapter, Gargarella distinguishes among four ways in which assemblies have addressed seemingly intractable issues: by the majority imposing its will, deferring the problem by dumping it on the future, by reaching a compromise based on an overlapping consensus, and by sweeping the conflict under the carpet by adopting mutually contradictory or ambiguous clauses. In addition to the Latin American cases he cites, the uneasy combination of regional autonomy and national unity in the 1978 Spanish constitution offers an illustration: “the price paid for such heavy compromise was that Spain, a country with relatively intense regional difficulties, was given a poor and incomplete territorial formula with the potential for worsening existing problems” (Bonime-Blanc Reference Bonime-Blanc1987, 89–90). Gargarella also notes the illiberal and dangerous tendency, recently observed in some post-Communist constitutions (Elster Reference Elster, Choudry and Ginsburg1993, 198–9), to limit the affirmation of rights by a blanket clause about the need to respect “public order and morality.” Postcolonial constitution makers in India retained similar restrictions from the preceding regime.

In her chapter, Lerner affirms, as noted earlier, that the success of constitution making in deeply divided societies depends on prior informal deliberations between the conflicting sides. As she notes, an “overlapping consensus” may not be feasible, if liberal constitutionalism is seen as representing one side of the conflict rather than a neutral ground (echoing the saying that universalism is the particularism of the rich). While joining Gargarella in citing deferral and ambiguity among the “solutions” to conflict, she adds the technique of including nonjusticiable or “aspirational” clauses in the constitution. Examples include the guarantees of social and economic rights that were inserted in some post-Communist constitutions to placate the former Communists (Elster Reference Elster, Choudry and Ginsburg1993, 198); and some “Directive Principles” in India’s constitution. In the post-Communist cases, though, the constitutional courts took these “parchment rights” more seriously than the non-Communist framers intended them to be.

Several chapters discuss assemblies that were simultaneously constituent and legislative. In their chapters, Bucur et al. and Negretto address the claim by Elster (Reference Elster, Bauman and Kahana2006) that such mixed assemblies have a tendency to adopt legislative-centric constitutions, as illustrated by the Polish assembly of 1921 and the French assembly of 1946. Bucur et al. confirm the tendency for the case of investiture rules in government formation in European assemblies. Negretto is more skeptical, citing his finding that in the context of Latin American presidential regimes the choice between a pure and a mixed assembly makes no difference for the allocation of power among the branches of government. Even assuming he is right, it would still make sense, in the study of a given mixed assembly, to explore the hypothesis that its decisions might have been shaped by institutional interest. The mechanism is intelligible and testable.

Independently of this effect, mixed assemblies are easily distracted from their constituent task by other issues. As Madison wrote in his preamble to the Virginia resolution calling for the Federal Convention, it would be “preferable to a discussion of the subject in Congress, where it might be too much interrupted by ordinary business” (Farrand Reference Farrand1966, Vol. III, 560). Such interruptions occurred regularly if unpredictably in the French Constituante, where in a speech from 1789 Robespierre (1912–67, Vol. VI, 118–20) cited them to argue against a proposal to set aside all but two days of the week for constitutional matters only. In her chapter, Malagodi observes that the dual role of the Pakistani constituent assembly “led to inordinate delays” and, as noted later, a lack of “the constitutional moment.” In times of transition, the dual role may be inevitable, but as we shall discuss shortly, legislative tasks can also be allocated to a separate body.

In some cases, constitutional conventions and legislative assemblies have operated side by side, as did for instance the Federal Convention and the Continental Congress. In his chapter, Negretto discusses the existence of parallel assemblies in Ecuador 1997–8 and 2007–8, in Colombia 1990–1, and in Venezuela 2007. As he writes, in a stark understatement, the conventions “had a difficult coexistence with the ordinary legislature.” As he also notes, in the Colombian case one of the problems was solved or mitigated by a compromise banning members of the convention from competing in the forthcoming legislative elections. This measure echoes that taken by the (mixed) French Constituante on May 16, 1791, when the framers declared themselves ineligible to the first ordinary legislature, to prevent suspicions that they were carving out a place for themselves in that assembly. In both cases, decisions were taken to maintain some degree of separation of constituent and legislative powers. Although the French decision proved disastrous – the first legislature, filled with inexperienced men, was easily dominated by the Jacobin clubs – it did address a genuine problem.

IV. Additional Questions

The present volume offers a number of case studies of failures of constitution making. Malagodi’s chapter is explicitly devoted to two such failures, in Pakistan and Nepal. Gylfason’s chapter details the mechanisms that have, so far at least, blocked the constituent process in Iceland. In her chapter, Lerner distinguishes among several mechanisms that led to failures of constitution making in Indonesia, Egypt, and Israel. Although she counts the process in Sri Lanka as a success, in the sense that it did lead to the adoption of a constitution, it might also count as a failure if the criterion for success is, as she also writes, the enactment of “an enduring democratic constitution” (our italics.) Finally, Elster’s chapter on 1814 describes the near-miraculous self-fulfilling wishful thinking that enabled the Norwegian framers to avoid a failure that to sober observers seemed inevitable.

In his chapter on the political psychology of constitution making, Elster suggests that one cause of such failures might be the absence of a crisis that can generate the will to overcome partisan and short-sighted interests. (In a phrase attributed to Winston Churchill, “Never let a good crisis go to waste.”) It is perhaps significant that the Icelandic process ran out of steam, as Gylfason explains in his chapter, when the economy improved. In her chapter, Malagodi suggests that the failures of constitution making in Nepal and Pakistan were due to delays and political infighting, leading to what she calls “an irrevocable loss of the constitutional moment.”

The foregoing discussions have been mainly oriented toward explanatory issues, although occasionally we have also touched on normative questions. One may ask, more generally, about the optimal design of a constituent assembly. One approach might be to search for the design most conducive to a good constitution. Another might be to search for the design that is least susceptible to distortion by normatively irrelevant factors, such as self-interest, group interest, institutional interest, cognitive bias, passion, and prejudice. Elster (Reference Elster2013, chapter 4) opts in the main for the second approach, while also emphasizing the desirability of constructive designs that will enable framers to determine where the shoe pinches and how to make good shoes. The first task requires broad representation of citizens’ interests, inducing a preference for proportional representation over majority voting in single-member districts. The second task might seem to require that delegates have some political experience, as was the case, for instance, of the American framers. As Gylfason explains in his chapter, the Icelandic selection of delegates excluded current MPs and cabinet ministers from the constitution-making body, while making mayors and other local politicians eligible. Although this proposal seemed attractive – mayors might be expected to know both where the shoe pinched and how to make better shoes – the twenty-five delegates included seven academics (two mathematicians!), five artists and persons associated with the media, as well as three lawyers. Three former MPs – but no local politicians – represented the political experience (complete list in Elster Reference Elster, Ingimundarson and Urfalino2016, 193–5). The paucity of accessible data makes explaining constitution making hard. Incomplete understanding results from insufficient documentation of plenary proceedings. James Madison’s notes from the Federal Convention are not only unusually complete, but also unusually reliable, in that they were not written for an external audience or for posterity (but see Bilder Reference Bilder2015 for some nuances). Elster’s chapter on 1814 relies on the letters by one framer, the History written by his son, the Recollections of another framer, the diary of the prince-regent, and on the official minutes. While extremely suggestive, these sources are also incomplete. In this respect, Gylfason’s chapter is in a class by itself, as he received the largest number of votes in the elections to the Constitutional Council. His chapter would have found a natural place in Constitution-Makers on Constitution-Making (Goldwin and Kaufman Reference Goldwin and Kaufman1988) if that volume were to be reedited. When there is a reasonably good record of the debates in the assemblies, supplemented and triangulated by letters and diaries of the framers, we can engage in the kind of micropolitical analysis pioneered by Ermakoff (Reference Ermakoff2015). Even in the best of cases, the reconstruction of the motives and beliefs of individuals, including their beliefs about the beliefs and motives of others, is an arduous and uncertain task. In his chapter on political psychology, Elster, following the French moralists as well as the framers at the Federal Convention, distinguishes among reason, passion (including prejudice), and interest as motivations of framers. In some cases, one may reasonably impute one of them to a given framer in a given situation, although the imputation is complicated by the fact that speakers, letter-writers and even diarists tend (like the rest of us) to profess noble motivations that they do not always harbor. Elster’s most robust claim is that earlier writers on constitution making have overemphasized the causal efficacy of “ideas” (or reason) and interest, while neglecting that of passion. Hence, he suggests, the classical metaphor that constitutions are documents written by Paul when sober to restrain Paul when drunk should be turned upside down: successful framers tend to be drunk, that is, filled with passion. In particular, he endorses Kant’s claim that without enthusiasm nothing great is ever achieved in history. Other emotions can, to be sure, undermine successful constitution making, in particular those generated by ethnic and especially religious conflicts.