Abstract Constituent conventions have traditionally occupied a privileged position within normative theories of constitution-making. It is claimed that compared to a constituent legislature, an assembly specially commissioned to write a constitution is better able to reflect citizens’ preferences and promote an impartial constitutional design. This article argues that the superiority of conventions over any type of constituent legislature has no support on either theoretical or empirical grounds. It also proposes that conventions may incur significant risks when a constitution is replaced within a democratic regime. Conventions often transgress their mandate by asserting their right to express or submit their decisions to the legally unbound power of the people. This claim to incarnate or represent the sovereign popular will increases the likelihood of conflicts between the convention and established institutions and, most crucially for the preservation of democracy, facilitates the use of this body by a dominant political group to legitimize its capture of state power. These arguments are supported with a critical analysis of the alleged virtues of conventions, a discussion of the history and theory of these bodies in the USA, and a comparative study of recent cases of constitutional replacement by conventions in Latin America. 1. Introduction Normative theories of constitution-making have assigned constituent conventions a privileged position among democratic constitution-making bodies. It is argued that compared to a constituent legislature, an assembly specially commissioned to write a constitution may better reflect citizens’ preferences and promote an impartial constitutional design. From this perspective, constituent conventions may help to create a more solid democratic foundation for new constitutions, contribute to a more balanced distribution of powers among state authorities, and ensure constitutional stability. I argue that the superiority of conventions over constituent legislatures, regardless of the selection method, composition, or decision-making procedures of each body, has no support on either theoretical or empirical grounds. I also propose that conventions may incur significant risks when a constitution is replaced within a democratic regime. In order to secure democratic stability, the task and powers of a constituent assembly must be executed within the limits of a pre-established legal framework. Yet conventions often transgress their mandate by asserting their right to express or submit their decisions to the legally unbound power of the people. This claim to incarnate or represent the sovereign popular will increases the likelihood of conflicts between the convention and established institutions and, most crucially for the preservation of democracy, facilitates the use of this body by a dominant political group to legitimize its capture of state power. I support these arguments with a critical analysis of the alleged virtues of conventions, a discussion of the history and theory of these bodies in the USA, and a comparative study of recent cases of constitutional replacement by conventions in Latin America. This article is organized as follows. Section 2 starts with a conceptual and comparative analysis of democratic constitution-making bodies. Section 3 assesses the relative benefits of constituent conventions vis-à-vis constituent legislatures. Section 4 discusses the problems involved in the concept of a legally limited convention in American constitutional law. Section 5 provides a comparative analysis of the use of constituent conventions in Latin American democracies from 1990 to 2010. A brief conclusion discusses the alternatives available to avoid the perils of a partisan convention when a constituent legislature is not appropriate to adopt a new democratic constitution. 2. Democratic constitution-making bodies A variety of collective bodies may be involved in a constitution-making process: constitutional commissions, executive bodies, round tables, national conferences, constituent conventions, and constituent legislatures. In democratic settings, however, the last two are the most common instances that work as assemblies where constitutional texts are deliberated, negotiated, and finally voted on.1 I will thus restrict my analysis to constituent conventions and legislatures, showing how they compare to each other in nature, frequency, and organizational form.2 Constituent conventions are assemblies created for the sole or primary purpose of adopting or proposing a new constitution.3 According to this definition, the key feature of these bodies is that they must be dissolved after the approval of a new constitutional text.4 Although conventions can also be used to adopt or propose revisions to the existing constitution, throughout this article I will only take into account their constituent function. Finally, in spite of the fact that convention delegates can be selected by various methods, I will focus mainly on elected conventions because at least since the twentieth century, popular election has been the most common form of selection of all constituent assemblies, particularly—but not only—in transitions to democracy and democratic regimes.5 The paradigmatic example of this body, which inspired much of the existing theorization around it, was the Philadelphia Convention of 1787. Similar cases can be found in the Norwegian constituent assembly of 1814, the Danish constituent assembly of 1848–1849, the German convention of 1948–1949, and the Portuguese constituent assembly of 1976. In general, however, this type of constituent assembly has been used in the history of relatively few countries or regions. From 1900 to 2015, only thirty-seven out of a sample of 124 deliberative bodies created to approve a new constitution may be classified as constituent conventions.6 Most of the assemblies of this type that have existed in the world were created in the American states. According to Hoar, the New Hampshire conventions of 1778 and 1781–1783 and the Massachusetts convention of 1780 initiated the convention movement in the USA.7 The practice proliferated throughout the nineteenth century and several states used these bodies during the twentieth century. Since the 1970s, however, there has been a sharp decline in the use of conventions for replacing or revising constitutions.8 Experts in American state constitutional law have counted 233 conventions between 1776 and 2005.9 At the national level, conventions have also been frequent in Latin America.10 Most of the first constitutions enacted after independence were made by assemblies that along with framing a new constitution were responsible for performing legislative and other government functions. The use of constituent conventions emerged over time, however, with the adoption of the 1830 Constitutions of Colombia and Venezuela, the 1853 Constitution of Argentina, and the 1870 Constitution of Paraguay. The use of these bodies continued well into the twentieth century and the first decade of the twenty-first century. From a total of eighty-three constitution-making processes that took place in this region from 1900 to 2014, twenty-six (31 percent) used a convention as constituent body.11 Conventions can come into existence irregularly or by virtue of a constitutional or legal authorization. Irregular conventions have been typical during revolutions, independence processes, and regime transitions, because in these contexts the preexisting legal order was suspended or no longer considered to be legitimate. In these cases, the authority of conventions may be derived from a special election or a political decision. Within an existing constitutional order, a convention can be irregular but may also be authorized by a constitutional provision or an ordinary law. Since at least the mid-nineteenth century, most conventions elected in the American states have been constitutionally authorized bodies.12 At the national level, however, most constitutions in the world lack provisions regulating the intervention of a convention to revise or replace their texts. Because of this legal vacuum only ten of a sample of thirty-seven conventions that existed between 1900 and 2015 were organized following a preexisting constitutional rule.13 In the absence of preexisting constitutional provisions regulating conventions, it is not surprising that few democratic regimes have used them to adopt a new constitution. As shown in Table 1, during the period from 1900 to 2015, twenty-five new constitutions have been adopted in the world in the context of a democratic regime at least five years old.14 Only five of these constitutions—all located in Latin America—were adopted by conventions. Within this group, the oldest democracies in which a convention was elected were those of Colombia and Venezuela, established in the late 1950s.15 In addition, in only two of these cases (Bolivia in 2009 and Ecuador in 1998) did conventions come into existence based on a preexisting constitutional or legal authorization. Table 1. Constitution-making bodies in democratic regimes, 1900–2015* Country Year Region Sub-region Constitutional/ legal Constituent body Denmark 1915 Europe Western Yes Constituent legislature Denmark 1953 Europe Western Yes Constituent legislature Finland 2000 Europe Western Yes Constituent legislature France 1958 Europe Western Yes Executive body Greece 1952 Europe Western No Constituent legislature Iceland 1944 Europe Western Yes Constituent legislature Ireland 1937 Europe Western Yes Constituent legislature Sweden 1974 Europe Western Yes Constituent legislature Switzerland 1999 Europe Western Yes Constituent legislature Hungary 2011 Europe Eastern Yes Constituent legislature Poland 1997 Europe Eastern Yes Constituent legislature Ukranie 1996 Europe Eastern Yes Constituent legislature Kenya 2010 Africa Sub-Saharan Yes Constituent legislature Thailand 1997 Asia East Yes Constituent legislature Sri Lanka 1972 Asia South No Constituent legislature Nepal 2015 Asia South Yes Constituent legislature Bolivia 2009 Latin America Andean Yes Constituent convention Colombia 1991 Latin America Andean No Constituent convention Ecuador 1998 Latin America Andean Yes Constituent convention Ecuador 2008 Latin America Andean No Constituent convention Uruguay 1952 Latin America South Yes Constituent legislature Uruguay 1967 Latin America South Yes Constituent legislature Venezuela 1999 Latin America Andean No Constituent convention Dominican Republic 2010 Caribbean South Yes Constituent legislature Trinidad & Tobago 1976 Caribbean South Yes Constituent legislature Country Year Region Sub-region Constitutional/ legal Constituent body Denmark 1915 Europe Western Yes Constituent legislature Denmark 1953 Europe Western Yes Constituent legislature Finland 2000 Europe Western Yes Constituent legislature France 1958 Europe Western Yes Executive body Greece 1952 Europe Western No Constituent legislature Iceland 1944 Europe Western Yes Constituent legislature Ireland 1937 Europe Western Yes Constituent legislature Sweden 1974 Europe Western Yes Constituent legislature Switzerland 1999 Europe Western Yes Constituent legislature Hungary 2011 Europe Eastern Yes Constituent legislature Poland 1997 Europe Eastern Yes Constituent legislature Ukranie 1996 Europe Eastern Yes Constituent legislature Kenya 2010 Africa Sub-Saharan Yes Constituent legislature Thailand 1997 Asia East Yes Constituent legislature Sri Lanka 1972 Asia South No Constituent legislature Nepal 2015 Asia South Yes Constituent legislature Bolivia 2009 Latin America Andean Yes Constituent convention Colombia 1991 Latin America Andean No Constituent convention Ecuador 1998 Latin America Andean Yes Constituent convention Ecuador 2008 Latin America Andean No Constituent convention Uruguay 1952 Latin America South Yes Constituent legislature Uruguay 1967 Latin America South Yes Constituent legislature Venezuela 1999 Latin America Andean No Constituent convention Dominican Republic 2010 Caribbean South Yes Constituent legislature Trinidad & Tobago 1976 Caribbean South Yes Constituent legislature * Independent states with populations of more than 300,000 that adopted a new constitution from 1900 to 2015 after at least five years since the first free and fair election in the country. Source: Author, based on Negretto, supra note 14; Boix, Miller, & Rosato, supra note 14, and various country sources. View Large Table 1. Constitution-making bodies in democratic regimes, 1900–2015* Country Year Region Sub-region Constitutional/ legal Constituent body Denmark 1915 Europe Western Yes Constituent legislature Denmark 1953 Europe Western Yes Constituent legislature Finland 2000 Europe Western Yes Constituent legislature France 1958 Europe Western Yes Executive body Greece 1952 Europe Western No Constituent legislature Iceland 1944 Europe Western Yes Constituent legislature Ireland 1937 Europe Western Yes Constituent legislature Sweden 1974 Europe Western Yes Constituent legislature Switzerland 1999 Europe Western Yes Constituent legislature Hungary 2011 Europe Eastern Yes Constituent legislature Poland 1997 Europe Eastern Yes Constituent legislature Ukranie 1996 Europe Eastern Yes Constituent legislature Kenya 2010 Africa Sub-Saharan Yes Constituent legislature Thailand 1997 Asia East Yes Constituent legislature Sri Lanka 1972 Asia South No Constituent legislature Nepal 2015 Asia South Yes Constituent legislature Bolivia 2009 Latin America Andean Yes Constituent convention Colombia 1991 Latin America Andean No Constituent convention Ecuador 1998 Latin America Andean Yes Constituent convention Ecuador 2008 Latin America Andean No Constituent convention Uruguay 1952 Latin America South Yes Constituent legislature Uruguay 1967 Latin America South Yes Constituent legislature Venezuela 1999 Latin America Andean No Constituent convention Dominican Republic 2010 Caribbean South Yes Constituent legislature Trinidad & Tobago 1976 Caribbean South Yes Constituent legislature Country Year Region Sub-region Constitutional/ legal Constituent body Denmark 1915 Europe Western Yes Constituent legislature Denmark 1953 Europe Western Yes Constituent legislature Finland 2000 Europe Western Yes Constituent legislature France 1958 Europe Western Yes Executive body Greece 1952 Europe Western No Constituent legislature Iceland 1944 Europe Western Yes Constituent legislature Ireland 1937 Europe Western Yes Constituent legislature Sweden 1974 Europe Western Yes Constituent legislature Switzerland 1999 Europe Western Yes Constituent legislature Hungary 2011 Europe Eastern Yes Constituent legislature Poland 1997 Europe Eastern Yes Constituent legislature Ukranie 1996 Europe Eastern Yes Constituent legislature Kenya 2010 Africa Sub-Saharan Yes Constituent legislature Thailand 1997 Asia East Yes Constituent legislature Sri Lanka 1972 Asia South No Constituent legislature Nepal 2015 Asia South Yes Constituent legislature Bolivia 2009 Latin America Andean Yes Constituent convention Colombia 1991 Latin America Andean No Constituent convention Ecuador 1998 Latin America Andean Yes Constituent convention Ecuador 2008 Latin America Andean No Constituent convention Uruguay 1952 Latin America South Yes Constituent legislature Uruguay 1967 Latin America South Yes Constituent legislature Venezuela 1999 Latin America Andean No Constituent convention Dominican Republic 2010 Caribbean South Yes Constituent legislature Trinidad & Tobago 1976 Caribbean South Yes Constituent legislature * Independent states with populations of more than 300,000 that adopted a new constitution from 1900 to 2015 after at least five years since the first free and fair election in the country. Source: Author, based on Negretto, supra note 14; Boix, Miller, & Rosato, supra note 14, and various country sources. View Large Constituent legislatures are assemblies that work as ordinary legislatures both during the drafting and after the approval of a new constitution. In contrast to conventions, they are the most common form of constituent body. From 1900 to 2015, sixty-nine of a sample of 124 deliberative bodies created to approve a new constitution qualify as constituent legislatures.16 Like conventions, constituent legislatures also differ from the point of view of the source of their constituent authority. Using this perspective, Jon Elster has distinguished between three different types of mixed constitution-making bodies: mandated constituent legislatures, self-created constituent legislatures, and self-created legislating assemblies.17 Although this classification is very useful, it demands a few adjustments. Mandated constituent legislatures are assemblies elected to enact a new constitution and pass ordinary legislation. They are frequently used in transitions to democracy where there is either no legislature at the beginning of the process or the existing one lacks legitimacy to assume a constituent function. Examples of these bodies can be found in the making of the 1975 Greek Constitution, the first 1946 French Constitution (rejected by voters), the 1931 Spanish Constitution, the 1992 Slovakian Constitution, and the 1946 and 1988 Brazilian Constitutions. Self-created constituent legislatures are ordinary legislatures that decide on their own authority to transform themselves into a constituent body. In regard to these bodies, Elster does not clarify on what basis a legislature decides to take on the task of writing a new constitution, except that it is not derived from an electoral mandate.18 To make the adjective “self-created” as precise as possible, however, the term should be restricted to legislative assemblies that assume a constituent task following a political decision made by the incumbent government or existing legislative parties without any form of authorization, either electoral or legal. These bodies have been common during revolutions and independence processes. For instance, legislatures self-appointed as constituent assemblies approved the 1776 constitutions of New Hampshire, South Carolina, and Virginia.19 We may also find this type of body in the adoption of some independence constitutions in Eastern Europe, such as those of Croatia in 1990 and Estonia in 1992.20 The drafting of a new constitution by a self-appointed constituent legislature also works as a non-legal but politically expedient method of replacing constitutions in authoritarian regimes.21 Constitution-making bodies of this type adopted most constitutions in Bolivia during the twentieth century and in Venezuela from 1904 to 1925. Self-created legislating assemblies are a hybrid type of mixed body, intermediate between constituent conventions and ordinary legislatures. These are constituent assemblies that were supposed to dissolve after enacting a new constitution but decided on their own authority to continue as ordinary legislatures. The Indian constituent assembly of 1946–1950 and the 1934 constituent assembly of Brazil may fit this category. Elster’s classification is not exhaustive, however. A missing category of mixed constituent bodies is what we may call constitutionally authorized constituent legislatures. These are ordinary legislatures that can turn themselves into constituent bodies following a procedure established in the existing constitution or legal instrument of constitutional status. This occurs when the existing reform procedure enables the ordinary legislature to replace the constitution in force. Constitutionally authorized constituent legislatures have been used in some transitions to democracy where the old constitution was amended or an interim constitution was created to authorize parliament to enact a new constitution.22 This mechanism has been particularly import- ant, however, in the adoption of new constitutions within established democratic orders. In particular, and as shown in Table 1, this type of legislature has been used to pass seventeen of the twenty-five constitutions enacted in democratic regimes from 1900 to 2015. There are different forms of regulating the adoption of a new constitution by the legislature. Some parliamentary constitutions enable the legislative assembly to adopt a new constitution because their revision procedure makes no distinction between piecemeal amendments and wholesale replacements. In 1953, the Danish parliament adopted a new constitution following the 1915 Constitutional Act, which established a uniform procedure of constitutional reform by the legislature.23 In 1974 the Swedish parliament replaced the 1809 Instrument of Government following the amendment process regulated within it.24 The Finnish parliament passed a new constitution in 2000 using a reform process described in the 1928 Constitutional Act.25 In all these cases, the procedure required a new parliamentary election before the legislature could approve the desired constitutional changes and, in the case of Denmark, also ratification of the reform in a popular referendum. A few separation of powers systems have also authorized the legislature to either amend or replace the constitution, sometimes distinguishing between these procedures. In the case of Uruguay, all constitutions since 1934 have authorized the legislature to propose and pass either partial or total reforms, with ratification in a referendum. This provided the basis for the adoption of the 1942, 1952, and 1967 constitutions.26 The system is similar in Switzerland, although in this country there is a clearer distinction between amendments and replacements. It was based on this distinction that the Swiss parliament adopted a new constitution and submitted it to popular and cantonal ratification in 1999.27 3. The alleged superiority of constituent conventions Normative theories of constitution-making have argued that compared to a constituent legislature, an assembly specially commissioned to write a constitution may enhance the democratic legitimacy of the constitution and promote an impartial design. In what follows I show that these arguments lack strong theoretical foundations or empirical support. In addition, I propose that there are significant risks involved in the use of conventions for the replacement of constitutions within a democratic regime. The original argument in favor of conventions is derived from the theory of constituent power. This theory, as formulated by key participants of the American and French revolutions such as James Wilson, Thomas Paine, Thomas Jefferson, and Sieyès, postulates that the right to create and replace constitutions belongs to the people, not to government bodies.28 Given this right, constitutions should be adopted or altered only by special assemblies independent of constituted authorities, in particular of those responsible for enacting ordinary laws.29 As Jefferson argued in his Notes on the State of Virginia, if the constitution is an act above the powers of the ordinary legislature, the legislature should not be allowed to alter constitutional provisions by its own decision. Otherwise, legislators would be judges in their own cause.30 Because their members are selected for the purpose of writing a new constitution, special conventions are also seen as producing documents with a high level of democratic legitimacy. Bruce Ackerman recommended using the American model of constituent convention in Eastern Europe to provide new constitutions in this region with a strong democratic foundation.31 More generally, Jon Elster has argued that “constitutions produced by conventions tend to have greater legitimacy and hence tend to enjoy greater stability.”32 In his view, conventions can be elected with a view to representing a variety of interests and programs rather than maximizing the representation of a particular group or ensuring stable governance, as might occur with a constituent legislature. A convention would also provide voters with the opportunity to elect delegates based on their reform proposals and not on other considerations.33 A very different type of normative argument in favor of conventions is epistemic in nature. According to Elster, these assemblies may also promote active and moral aptitude in the framers by inducing full attention and concentration on the task they were commissioned to do and by reducing the impact of institutional interests on their decisions.34 Unlike constituent legislatures, which take responsibility for legislative and constitutional decisions at the same time, constituent conventions enable the framers to hold sustained and focused deliberations on constitutional issues alone.35 In addition, members of a constituent legislature are more likely than delegates to a convention to be biased toward the legislature in the design of the machinery of government. Just as an executive constitution-making body would write an important role for itself in the constitution, so a constituent legislature would give a preponderant importance to the legislative branch at the expense of the executive and the judiciary.36 From a normative point of view about the design of a democratic constitution-making process, it certainly makes sense to choose a constituent body that gives more weight to the preferences of citizens than to those of state authorities and creates a sense of collective ownership over the new text. It is also defensible from an epistemic perspective to select a constituent assembly that removes cognitive biases and self-interest. It is not clear, however, that a constituent convention has an inherent advantage over any type of constituent legislature on these grounds. From a comparative perspective, the constituent power theory has not been adopted in several legal systems that enable the legislature to adopt a new constitution. The influence of the theory has also varied by historical period and region of the world. Nevertheless, the notion that fundamental constitutional changes should only come from the people as the sovereign authority is still very strong in constitutional and political theory. For this reason, it is important to start by considering the idea that the use of conventions is a necessary implication of the distinction between constituent and constituted powers and the superiority of the former over the latter. The principle of separation between constituent and constituted powers demands that legislators should act within the limits of the powers that the people delegated to them under the constitution. This clearly implies that they cannot alter those powers on their own authority. But why could legislators not have a constituent role if they are explicitly authorized by voters to approve a new constitution? Moreover, popular ratification of the changes may also be required so that legislators’ margin of autonomous decision is even more restricted. In the end, both constituent conventions and legislatures are representative bodies. So, unless one thinks that only the election of a convention counts as an authentic expression of the constituent power of the people, there is no reason to think that constitutions should always be made by special conventions. The argument that a constitution adopted by a popularly elected convention has a stronger claim to democratic legitimacy than one produced by a legislature is obviously correct if we think of a self-appointed constituent legislature. A legislature that arrogated to itself the right to create a constitution without an electoral mandate or constitutional authorization has no basis from a democratic perspective to sustain that right. Yet it is not clear why a constitutional text approved by a legislature elected to have an initial constituent function or authorized to assume this role after an intervening democratic election would be any less legitimate than a popularly elected convention. In the case of an assembly initially elected to work as legislative and constituent body, one may argue that this dual role would lead voters to cast their votes based on considerations alien to the content of the new constitution, such as general party platforms or the personal traits of a candidate. A similar set of motivations, however, may explain voters’ preferences when—as is often the case—those who compete for a position in a convention do not yet have well-formed preferences about their own reform proposals during the election. When the legislature is allowed to assume a constituent function only after an intervening election (as is the case with some constitutionally authorized constituent legislatures), the idea of a democratic deficit is even less persuasive. In this situation, the public debates about constitutional reform that preceded the new election of the legislature should exert a predominant influence on voter’s choices. It is true that unlike legislators, who are usually elected on a partisan basis, delegates to conventions may be partially elected, elected as independents, appointed, or randomly selected. This range of options enables conventions to enhance representation by incorporating ordinary citizens and traditionally excluded groups. In most democratic contexts, however, constituent conventions are elected on a partisan basis, just like constituent legislatures.37 In this situation, the degree to which an assembly represents the plurality of political interests and views present in society hinges on the proportionality of the system under which its members are elected. Whether these interests and views have an influence on final decisions also depends on how consensual the decision rule is. From this perspective, a constituent legislature that is elected by a proportional formula and makes decisions by qualified majority would be more inclusive than a convention that is elected by plurality and makes decisions by simple majority. In actual practice, constituent conventions do not have a particular advantage over constituent legislatures in terms of the inclusiveness of the electoral rules used to select their members. Most of the constitution-making bodies elected in the world since 1900, whether special conventions or constituent legislatures, have been elected by proportional formulas.38 In regard to decision rules, although qualified majority thresholds are generally rare they are more frequently observed in constituent legislatures than in conventions.39 There may be a theoretical reason for this practice. As Ulrich Preuss has argued, since constituent conventions incarnate the constituent power of the people, they should make decisions by majority because it is the only rule that secures the principle of equality.40 The proposal that conventions promote the active aptitude of framers to focus on constitutional issues alone is the most persuasive argument in favor of these bodies. Reformers who must divide their time between designing a constitution and making ordinary legislative decisions may not only confuse the two tasks, writing into the constitution matters that are better left to statute, but also have less time to concentrate on the discussion of constitutional provisions. This may detract from the quality of constitutional design. However, mixed assemblies may also achieve a relatively efficient division of labor between lawmaking and constituent activities. Just as ordinary parliaments have a committee system to promote the specialization of legislators on certain matters, nothing prevents a constituent legislature from creating an internal division of labor so that a subgroup of the assembly concentrates on the drafting of the constitution. For instance, during the making of the 1961 Venezuelan Constitution a broadly representative bicameral commission was responsible for submitting the draft of the new constitution to the plenary of Congress.41 The idea that a legislature would be tempted to benefit itself at the expense of other branches of government is based on the assumption that the interests of existing institutions are primarily articulated through the organizational form of the constituent body. Yet democratic representation in both constituent conventions and constituent legislatures is normally channeled through political parties. This means that the institutional preferences of constitution-makers are more likely to be shaped by the concrete interests of their parties than by the abstract interest of the collective body in which they gather as representatives of the people. As politicians, constitution-makers tend to defend the institutional interests of their parties because doing so benefits them individually, by helping them win office and have influence over important decisions.42 For this reason, reformers who have a partisan link with the legislature or the executive are prone to making constitutional choices that favor these branches regardless of whether the constituent body is a legislature or a convention. If the interests of existing institutions are represented through the parties that control or expect to control them, one way to induce impartiality in constitution-making would be to forbid parties from participating in the constituent body. This would be the case of assemblies made up of randomly selected citizens or delegates elected on a nonpartisan basis. This solution, however, would imply that removing the influence of group and institutional interests does not depend on the type of body per se but on its composition. Another possibility is to allow political parties to field candidates for the election of delegates but to postpone the implementation of the new constitution until sometime after it was adopted. This would increase constitution-makers’ level of uncertainty as to which institutions would benefit them most, thus inducing a more impartial constitutional design.43 However, if this impartial perspective effectively takes place it would derive from the delay in implementing the constitution, not from the nature of the constituent body. Although many arguments in support of conventions are based on principle and thus not subject to empirical validation, some make implicit or explicit causal claims that can be observed. For instance, if constitutions adopted by constituent conventions have a stronger democratic foundation than those written by constituent legislatures, then the former may last longer than the latter. More explicitly, if constituent legislatures are prone to giving undue influence to legislative interests, constitutions produced by them should invest the executive and the judicial branch with less power than a constituent convention would. There are several methodological problems in testing these claims, the most important of which is the large number of confounding variables potentially involved in measuring the causal effects of procedural rules. The empirical analyses performed so far, however, do not seem to validate the hypothesized impact of different constituent bodies. To date, there is no cross-regional test of the durability of constitutions adopted by different constituent bodies. Yet the finding of recent empirical works that more detailed constitutions survive longer suggests the plausible hypothesis that legislatures (at least in regard to the durability of constitutions) might be better constituent bodies than special conventions.44 Members of organized groups are likely to have stronger influence on members of a permanent legislature than on delegates to a temporary assembly. Being concerned with reelection, legislators are also likely to pay more attention to the interests of these groups than are delegates of a special convention. If this assumption is correct, then legislators may be more inclined to write longer and more detailed provisions into the constitution to show their constituents that their interests have been protected. These constituents, in turn, would have more incentives to support the constitution in the long run. In a more direct analysis of the impact of the constituent body on constitutional durability, but looking only at Latin American cases, it has been found that there is no significant difference between the effects of constituent conventions or legislatures on the lifespan of constitutions.45 Whether constituent legislatures engage in self-dealing has been tested by Ginsburg, Elkins, and Blount, based on a sample of 411 episodes of constitution-making around the world from 1789 to 2005.46 Correlating the use of pure and mixed constituent assemblies with an index of parliamentary powers, they conclude that there is no evidence to sustain the claim that constituent legislatures are more likely to strengthen the powers of the legislature. These results may be questionable, however. The sample used by these authors includes both democratic and authoritarian constitution-making episodes, so that the units of observation are too heterogeneous to make a reliable comparison possible.47 The same finding has been reported, however, in tests restricted to constitutions made under competitive conditions.48 To sum up, the superiority of constituent conventions over constituent legislatures, regardless of the election, composition, or decision-making procedures of each body, has no support on theoretical or empirical grounds. A constituent legislature explicitly authorized by voters to pass fundamental constitutional changes can have the same democratic credentials as an elected convention and both can be equally representative if they are elected by inclusive electoral rules or make decisions by qualified majority. A constituent legislature with an internal division of labor may also promote an adequate level of specialization on constituent tasks. As for the influence of group and institutional interests on design, the typical partisan convention is not more immune to them than a constituent legislature is. In addition to the fact that constituent conventions have no inherent advantages, the usual convention made up of political parties may incur significant risks when a new constitution is adopted within an existing democratic regime. Whatever else reformers may want to achieve through the creation of a new constitution in a democratic order, it is crucial that the process not have any features that might jeopardize the stability and continuity of the political regime. This requires that the constitution-making body be able to coexist peacefully with the existing constituted powers and that its mission and powers be subject to the law. Constituent conventions are prone to violating these requirements. A special convention is likely to enter into several types of conflict with the ordinary legislature. Some of these conflicts may refer to minor jurisdictional issues. A convention, for instance, may decide to modify the deadline the legislature imposed to approve the new constitution. Yet other conflicts could revolve around whether the convention has sovereign, legally unlimited power. If a statute regulates the tasks and powers of the convention, it may reject these limitations based on the idea that the legislature is politically inferior to the convention. A dominant political group could use this conflict to justify its usurpation of legislative functions or interference with the judiciary.49 To be sure, a legislature dominated by a single party or coalition can also act arbitrarily or transgress the constitution. Yet a legislative assembly cannot easily legitimize those actions by claiming to be outside and above the existing constitutional order, as conventions often do. One possible way to minimize these risks in a democratic regime would be by making the convention a regular organ for the replacement of the constitution. This means, at a minimum, establishing in the constitution the conditions and procedure under which a convention is called, but it may also include rules about its election, powers, and decision-making process. The key objective of this regulation is to turn the convention into an institution regulated by and subordinated to the existing constitution and without jurisdiction over other state powers. Although most constitutions in the world do not provide for the election of a special convention for their replacement, that regulation could in principle be created. There is no better way to explore this issue than by analyzing the history of constitution making in the USA, where the concept of a legally limited constituent assembly originated. 4. Conventions in the American legal tradition In the American legal tradition the term “constitutional convention” refers both to the specialized task of this assembly—revising or replacing the constitution—and to the constitutional or legal limits observed in the execution of its mandate.50 However, as the following discussion makes evident, a close analysis of the use of conventions in the USA at the state and federal levels reveals that the idea of a legally limited convention is deeply problematic in both theory and practice. There are some issues on which all scholars on the use of conventions in American constitutional law seem to agree. One of them is that a convention is “constitutional,” as opposed to irregular, illegal, or revolutionary, if it is called by the existing legislature following explicit constitutional provisions.51 There also seems to be a consensus that whenever the constitution is silent about revision procedures, a convention would still be legal if the legislature enables its election and regulates its operation through statutory law.52 Based on these concepts, one may distinguish between three types of conventions: constitutional, extra-constitutional but legal, and extra-constitutional and illegal.53 In other words, only a convention called in the absence of constitutional provisions and without the consent of the existing legislature would be outside the existing legal order. The decisions of this convention could only become valid if a revolution eventually succeeds in imposing the new legality by force. The archetypal example of an illegal or revolutionary convention is that of the 1841 “People’s Convention” of Rhode Island.54 In the absence of an amendment procedure in the state’s colonial charter and in reaction to the government’s opposition to revising the constitution to expand the franchise, members of the so-called suffrage movement called a convention commissioned to enact a new constitution. The assembly was elected in an irregular election and the constitution it enacted was ratified in an unofficial referendum. A majority of the total number of adult male residents of the state supported the constitution, but many of those who voted for it were not qualified to vote under the existing laws. Although new executive and legislative authorities were elected under the new constitution, the rebel government failed to impose itself. The federal government sided with the Charter government, and members of the suffrage movement were put in jail or forced to leave the state.55 Although it did not rule on the merits of the case, the US Supreme Court implicitly acknowledged the illegality of the People’s Convention based on the recognition by the courts of Rhode Island and the president of the USA of the Charter government as the only legal authority in the state.56 Beyond this case, however, the legal foundation of a wide variety of conventions and convention acts is uncertain. Can a convention be legal if it is convened by the legislature against explicit constitutional provisions regulating a different mode of revision? The 1789 Pennsylvania Convention, the 1791 Delaware Convention, and the 1850 Maryland Convention were all called by the legislature in transgression of the existing revision procedures.57 Some scholars regard these bodies as irregular and illegal because in their view the legislature had no power to contravene the constitution.58 Others, however, argue that these conventions are extra-constitutional but nonetheless legal because the electorate supported them either before the legislature decided to call the convention or at the time of voting for its delegates.59 Given these contradictory interpretations of the same episodes, one is forced to conclude that the history of conventions in American constitutional law does not provide unambiguous guidance as to the use of conventions outside the revision procedures in force. Although this might seem a technical matter, at its root the problem revolves around an important normative question: where does the authority of conventions come from? If their authority originates in the constitution or, in case of no regulation, in a legislative act, then popular approval (even in a regularly called election) cannot validate a transgression of the legal framework. If, however, the authority of conventions comes directly from the people speaking through the electorate, then it does not matter whether the constitution is silent or establishes a different procedure: a convention would still be legal if it is approved by voters. Neither the practice nor the jurisprudence on conventions in the USA provides decisive support to either theory.60 Yet the position one takes on this issue is crucial because it will also determine how we assess the case of a convention that after being called regularly and according to existing procedures makes decisions that violate the legal limits imposed on its commission. Most state conventions elected since the nineteenth century have submitted their proposals of reform to popular approval.61 Some authors even think that conventions have the duty to ratify their revisions in a referendum if the constitution or the law is silent about it.62 Consider now the effect of legal transgressions of the convention in the light of popular ratification. If the authority of conventions comes directly from the people speaking through the electorate, then any violation of the legal limits imposed on their tasks can be validated ex post if the voters ratify the decisions of the convention in a referendum. These limits may refer to its commission to amend the constitution rather than to frame a new one, the time frame of its work, its decision-making procedures, or the prohibition on engaging in ordinary lawmaking.63 There is no better example of the ambiguity of the subjection of constituent conventions to the law than the actions of the Federal Convention of 1787. This convention produced a new constitution instead of amending the Articles of the Confederation as mandated. It also violated the existing amendment process by requiring ratification by nine instead of thirteen states and replacing state legislatures by popularly elected conventions as ratification bodies. Although the Continental Congress and state legislatures implicitly acquiesced to these actions, it was apparent that the convention exceeded its mandate. Madison justified the convention with two central arguments. The first was that since the people are unable to act spontaneously to alter or abolish their government when they deem it necessary, great constitutional transformations usually depend on irregular and unauthorized propositions made by a group of representatives. The second argument was that these actions entail no risk as long as “the people themselves” are the ones who decide whether to approve the proposed changes.64 Popular ratification, in other words, removes any preceding errors or irregularities. The fact that the Federal Convention invoked the revolutionary right of the people to alter or abolish their governments but nevertheless sought to act with at least the implicit consent of existing federal and local governments has led American constitutional theorists to disagree on the nature of this assembly. Whereas some authors (e.g., Kay) have labeled the decisions of the convention as plainly illegal and others (Amar) considered them legal, still others (Ackerman and Kaytal) view the strategy of the convention as “unconventional.” In other words, there is no consensus as to whether the Federal Convention was legal, revolutionary, or something in between.65 It is common to draw a contrast between the 1789–1791 Assemblée Constituante in France and the 1787 Federal Convention in America as two different models of constituent assemblies. According to this view, not only did the French assembly exercise constituent and legislative functions, but it also claimed to be “sovereign” and above all previous institutions. In contrast, the American assembly only drafted a new constitutional text and acted within, or at least not openly against, the framework of existing laws.66 However, the distinction between a French “legally unbound” model, and an American “legally limited” type of constituent assembly is blurred if the latter is assumed to be free to do anything, even to break the law, as long as the sovereign people ratifies its decisions ex post. The potentially arbitrary power of a convention explains why Madison had little faith in reproducing the Philadelphia experiment in the future and considered Jefferson’s idea of using periodical conventions as a risky method for altering the Constitution or resolving constitutional controversies.67 It also explains why the convention mechanism mentioned in Article V of the American Constitution has never been used at the federal level. As several scholars have observed, it is not only unclear how but whether this assembly could be legally limited.68 It is true that with the exception of the secession and reconstruction periods, most conventions elected in the American states after 1787 were called regularly and did not claim sovereign powers to encroach on or usurp the functions of constituted authorities.69 However, two circumstances peculiar to the American states account for this result. The first is that over time most state constitutions regulated the convention mechanism of revision. In spite of the ambiguities discussed above about the convention’s source of authority, a tradition of legal regulation provided state courts with some guidance as to how to decide on constitutional controversies regarding the powers and decisions of these bodies. The second reason is that while a state convention might claim a direct relation to the sovereign people of the state and demand freedom from legal regulation by the legislature, it still remains subordinated to the federal constitution and federal authorities. In other words, the existence of a federal government above state governments provides a check and an effective limit on what state conventions can do. 5. Conventions in contemporary Latin American democracies Unlike American state constitutions but similar to the American federal Constitution and the position of Madison on this issue, most national constitutions do not have provisions regulating their own replacement. Those that do include such provisions usually authorize the legislature, not a special convention, to replace them. If we add this lack of regulation to the uncertainties surrounding a special convention working in parallel with an ordinary legislature, it is only natural (as previously shown in Table 1) that most democratic regimes in the world have rejected the convention mechanism for adopting a new constitution. The fact that only Latin American democracies have used national conventions to replace their constitutions and that they came into existence in various ways makes the analysis of these cases particularly useful for shedding light on the nature of conventions and the conditions under which they can undermine democratic institutions.70 As I will show, most Latin American conventions entered into conflict with the legislature and claimed to have powers beyond the limits of their commission. However, whether they effectively jeopardized the stability or continuity of democracy depended on the level of partisan conflict around the organization and goals of the constitution-making process and, crucially, on the distribution of partisan power within the convention. The conventions that adopted new constitutions in Latin American democratic regimes fit the categories of constitutional, extra-constitutional but legal, and extra-constitutional and illegal assemblies discussed above in the context of the theory and jurisprudence of conventions in American constitutional law. Table 2 summarizes the sources of regulation of the conventions, their initial commission, and legal status in the five episodes of democratic constitution-making that took place between 1990 and 2010.71 Table 2. Regulation, mandate, and legal status of Latin American conventions, 1990–2010 Cases Source of regulation Mandate Legal status Bolivia (2004–2009) Constitution/Congress Draft constitution Constitutional Ecuador (1997–1998) Executive/Congress Draft constitution Extra-constitutional but legal Colombia (1990–1991) Executive/inter-party agreement Draft constitution* Extra-constitutional and illegal Ecuador (2007–2008) Executive Draft constitution Extra-constitutional and illegal Venezuela (1998–1999) Executive Draft constitution Extra-constitutional and illegal Cases Source of regulation Mandate Legal status Bolivia (2004–2009) Constitution/Congress Draft constitution Constitutional Ecuador (1997–1998) Executive/Congress Draft constitution Extra-constitutional but legal Colombia (1990–1991) Executive/inter-party agreement Draft constitution* Extra-constitutional and illegal Ecuador (2007–2008) Executive Draft constitution Extra-constitutional and illegal Venezuela (1998–1999) Executive Draft constitution Extra-constitutional and illegal *The Supreme Court invalidated the initial restriction to adopt specific reforms and enabled the convention to have full powers to decide on the content of the new constitution. View Large Table 2. Regulation, mandate, and legal status of Latin American conventions, 1990–2010 Cases Source of regulation Mandate Legal status Bolivia (2004–2009) Constitution/Congress Draft constitution Constitutional Ecuador (1997–1998) Executive/Congress Draft constitution Extra-constitutional but legal Colombia (1990–1991) Executive/inter-party agreement Draft constitution* Extra-constitutional and illegal Ecuador (2007–2008) Executive Draft constitution Extra-constitutional and illegal Venezuela (1998–1999) Executive Draft constitution Extra-constitutional and illegal Cases Source of regulation Mandate Legal status Bolivia (2004–2009) Constitution/Congress Draft constitution Constitutional Ecuador (1997–1998) Executive/Congress Draft constitution Extra-constitutional but legal Colombia (1990–1991) Executive/inter-party agreement Draft constitution* Extra-constitutional and illegal Ecuador (2007–2008) Executive Draft constitution Extra-constitutional and illegal Venezuela (1998–1999) Executive Draft constitution Extra-constitutional and illegal *The Supreme Court invalidated the initial restriction to adopt specific reforms and enabled the convention to have full powers to decide on the content of the new constitution. View Large In Bolivia a constituent convention was called under the constitution in force after the constitution was amended to include such a procedure. The 1967 Bolivian Constitution only regulated partial revision of the constitution through amendments whose necessity had to be first declared in Congress and then passed in a subsequent legislature, in both cases by two-thirds of the members present and voting in the session.72 In the midst of escalating social mobilization in favor of replacing the old constitution, it was reformed in February 2004 to allow Congress to convene a constituent assembly and regulate its internal procedures. Based on this reform, in 2006 Congress passed a law regulating the election of constituent assembly delegates; the decision-making process of the assembly, which required a two-thirds majority to pass the constitution; the relationship between the constituent assembly and the Congress; and final ratification of the constitution by referendum.73 The 1997–1998 constituent assembly of Ecuador serves as an example of an extra-constitutional but legal convention. The 1978 Ecuadorean Constitution did not have a procedure for its replacement. However, after congressional removal of the incumbent president in 1997, an interim president, using constitutional powers and with the agreement of Congress, convened a referendum asking for authorization to elect a constituent assembly.74 As a result of popular support obtained in the referendum, Congress passed a transitory constitutional provision to regulate the election and tasks of the constituent assembly. The remaining conventions were both extra-constitutional and illegal. Even within this category, however, one can observe two different alternatives. Colombia illustrates a consensual option, because in spite of its illegality the convention was regulated via a political agreement made between the government and opposition parties. After an unofficial referendum in the March 1990 congressional election provided support for the election of a constituent convention, President Barco issued an emergency decree calling a new, but this time official, referendum in the May presidential election. As this referendum again supported the election of the constituent convention, on August 2, 1990, the president-elect, Cesar Gaviria of the Liberal Party, signed an agreement with the leaders of the main political forces on the procedures by which the constituent convention would be elected and the aspects of the constitution that should be reformed.75 Based on this agreement, Gaviria used his state of siege powers to issue a decree (of very dubious legality) that stipulated the mode of election, rules of operation, and areas of reform that the convention could consider.76 This decree attempted to impose on the convention the content of the reforms it should adopt. In reviewing the decree, however, the Colombian Supreme Court upheld its legality with respect to the mode of election, time frame, and composition of the assembly but declared the convention free to decide on the content of the new constitution.77 Venezuela and Ecuador represent cases of a legal break by the executive without the consent of the legislature or a political negotiation with opposition parties. In both cases, the president unilaterally convened a referendum—without constitutional authorization or forcing the interpretation of existing rules—so that citizens would vote on whether an elected constituent convention should replace the constitution. Following this authorization, an assembly elected under the rules decided by the executive passed a new constitution subject to final ratification in a referendum.78 Note that I have codified the legal status of conventions based only on the initial source of regulation. It is important to note, however, that when considering particular steps of the process the use of terms “legal” or “illegal” may be subject to dispute. For instance, when the original constitution lacked an established procedure for calling a constituent convention, the election of this body often twisted existing legal rules. In the case of Bolivia, the possibility of replacement was added to the constitution in violation of the amendment procedure in force.79 In Ecuador, the interim government that organized the process in 1997 was not even the rightful successor of the government elected in 1996. On the other hand, in the cases of radical legal break, the government always intended to justify some aspects of the process on juridical grounds. In Venezuela, the referendum that the president convened to authorize the election of the constituent convention was based on a law enacted in 1997, and the Supreme Court validated its use.80 Even in Ecuador in 2007, where the process to convene the referendum to authorize the election of the convention was patently illegal, the Supreme Electoral Tribunal, which sided with the government, tried to provide some legal rationale for it. In spite of the different ways in which they were called into existence from the point of view of legality, all conventions entered into conflicts of various levels of intensity with the legislature. At the same time, although most of these conventions attempted to transgress the limits of their commissions, only two openly usurped legislative or judicial functions. As Table 3 illustrates, the number of parties with control over the decision rule was clearly associated with this outcome. Table 3. Partisan control, inter-institutional conflicts, and mandate execution of Latin American conventions, 1990–2010 Cases Partisan control* Attempt to transgress mandate? Convention– legislatureconflict? Interference with legislative or judicial functions? Bolivia (2004–2009) Two parties Yes Yes (high) No Ecuador (1997–1998) Three parties No Yes (low) No Colombia (1990–1991) Two parties Yes Yes (medium) Partial (1) Ecuador (2007–2008) One party Yes Yes (high) Yes (2) Venezuela (1998–1999) One party Yes Yes (high) Yes (3) Cases Partisan control* Attempt to transgress mandate? Convention– legislatureconflict? Interference with legislative or judicial functions? Bolivia (2004–2009) Two parties Yes Yes (high) No Ecuador (1997–1998) Three parties No Yes (low) No Colombia (1990–1991) Two parties Yes Yes (medium) Partial (1) Ecuador (2007–2008) One party Yes Yes (high) Yes (2) Venezuela (1998–1999) One party Yes Yes (high) Yes (3) * Minimum number of parties needed to approve the new constitution according to the decision rule of the convention. (1) Drafted the constitution but also decided to call early congressional elections. (2) Drafted the constitution but also assumed legislative functions, and claimed discretion to intervene in the affairs of the other branches of state power. (3) Drafted the constitution but also interfered in the activities of Congress and the judiciary. View Large Table 3. Partisan control, inter-institutional conflicts, and mandate execution of Latin American conventions, 1990–2010 Cases Partisan control* Attempt to transgress mandate? Convention– legislatureconflict? Interference with legislative or judicial functions? Bolivia (2004–2009) Two parties Yes Yes (high) No Ecuador (1997–1998) Three parties No Yes (low) No Colombia (1990–1991) Two parties Yes Yes (medium) Partial (1) Ecuador (2007–2008) One party Yes Yes (high) Yes (2) Venezuela (1998–1999) One party Yes Yes (high) Yes (3) Cases Partisan control* Attempt to transgress mandate? Convention– legislatureconflict? Interference with legislative or judicial functions? Bolivia (2004–2009) Two parties Yes Yes (high) No Ecuador (1997–1998) Three parties No Yes (low) No Colombia (1990–1991) Two parties Yes Yes (medium) Partial (1) Ecuador (2007–2008) One party Yes Yes (high) Yes (2) Venezuela (1998–1999) One party Yes Yes (high) Yes (3) * Minimum number of parties needed to approve the new constitution according to the decision rule of the convention. (1) Drafted the constitution but also decided to call early congressional elections. (2) Drafted the constitution but also assumed legislative functions, and claimed discretion to intervene in the affairs of the other branches of state power. (3) Drafted the constitution but also interfered in the activities of Congress and the judiciary. View Large In Bolivia, most parties agreed on the need to call a convention but disagreed profoundly on how to regulate the assembly and on the content of the future constitution. In this context, making the convention sovereign and “above” constituted powers was one of the government party’s central demands.81 Since this party won a majority of delegates in the convention but was unable to reach the two-thirds required by the congressional law that called it into existence, the convention’s claim to have sovereign power was meant to be used by the government to violate that requirement and adopt the constitution by majority rule.82 Several conflicts took place between the convention and Congress and between government and opposition around this issue.83 In the end, however, the government party (which also lacked a majority in the upper chamber of the bicameral legislature) backtracked from its attempt and the convention did not usurp or interfere with congressional functions. Moreover, the final text was negotiated with opposition forces in Congress after the government party failed to impose its own draft.84 Both the need for a convention and the procedure to call it into existence were widely agreed on among the major parties in the case of the 1997–1998 constitution-making process in Ecuador. The main parties also agreed on the general content of the future constitution and none of them won enough seats in the assembly to dictate decisions without support from the others.85 This explains the relatively peaceful coexistence between the legislature and the convention in spite of some conflicts between these two institutions regarding the term limits imposed on the writing of the new constitution. It also explains why in the execution of its mandate the assembly never claimed powers beyond its commission to draft a new constitutional text. In Colombia there was an initial inclusive agreement on calling the convention as well as on the general content of reforms. In addition, no single party or coalition won a majority of seats in the convention. The Supreme Court ruling, that declared invalid the decree imposing on the convention the content of the reforms it should adopt, based the decision on the idea that the authority of the convention came directly from the nation, as the holder of popular sovereignty.86 Although this did not imply that the convention could interfere with the normal functioning of constituted powers, opposition parties in the assembly nevertheless claimed that this body had the power to call for the anticipated election of a new congress. The main reason for this move was that opposition parties had increased their support in recent elections and expected to repeat that result in a new congressional election.87 The initiative to end the terms of legislators elected in 1990 created a severe conflict between the two bodies, which could only be solved after a compromise was reached between the government and opposition parties. The compromise accepted the right of the assembly to call for the election of a new congress after the approval of the constitution, provided that members of the convention did not compete in this election.88 A very different structure of conflicts and balance of forces characterized the cases of conventions called by unilateral executive decision. Although few parties rejected the calling of a convention, in both Venezuela and Ecuador (2007) the procedure to call a convention, the nature of this body, and the content of the new constitution were highly controversial. In the case of Venezuela, one of the provisions issued by the national electoral commission, to be included in the referendum to authorize the calling of the convention, declared this body to be sovereign. Ruling on the constitutionality of these provisions on April 1999, however, the Supreme Court declared the sovereignty clause invalid.89 In spite of this, one of the first acts of the convention after the government party won an overwhelming majority in the election of delegates was to declare itself above the constituted powers. Following this declaration, it intervened in the judiciary and restricted the activities of Congress. 90 In Ecuador between 2007 and 2008, a government-dominated convention also claimed sovereign powers to absorb legislative functions even though it lacked a mandate to do so. Article 1 of the executive decree that called this convention into existence stated that the convention had “full powers” (plenos poderes) to adopt a new constitution. Yet it also stated that any change that the assembly introduced in the institutional structure of the state could only become effective after the new constitutional text was approved.91 Once elected, however, the assembly passed a resolution asserting that its powers were above those of any of the existing branches of government, including Congress and the judiciary. Based on this decision, it explicitly assumed the power to legislate and declared the existing congress to be in recess.92 In sum, regardless of their origin, most conventions had a difficult coexistence with the ordinary legislature and claimed to be “sovereign” in the double sense of being free to decide on the content of the new constitution and above constituted powers. Whether they succeeded in their attempt, however, was related to the level of partisan conflict around the organization and goals of the process and depended ultimately on the balance of political forces in the assembly. Whenever a party or coalition had the resources to manipulate the convention to their own advantage, they did so using the purportedly superior democratic legitimacy of this body to justify their seizure of state power. These processes have not been neutral for democracy. Political conflicts associated with the coexistence of special conventions and legislatures were in most cases only temporarily disruptive, although in one case (Bolivia) led to high levels of political confrontation and violence that arguably affected the quality of the democratic regime.93 Some of these conflicts, however, had a more negative and lasting political effect when a dominant political force used them strategically to usurp legislative and other government functions, as in Venezuela and Ecuador (2008). In these two cases, the country ceased to be democratic at the time or a few years after adopting the new constitution.94 6. Conclusion This article has argued that constituent conventions are not inherently superior to all forms of constituent legislatures, whether from the point of view of democratic legitimacy, constitutional design, or rational decision-making. It has also proposed that conventions may entail significant political risks when used to replace a constitution within an established democratic regime. Claiming to express the will of the people, conventions tend to enter into conflict with the ordinary legislature and, more consequentially for the continuity of democracy, may be used by a partisan majority to legitimize acting beyond their commission. To be sure, there clearly are circumstances in which calling a constituent convention may be appropriate. This would be the case when the lack of legitimacy or suboptimal performance of the existing institutions and the demand for a new constitution are closely associated with the way the legislature works. In this situation, since the legislature itself needs to be reformed, an independent convention may be more likely to enhance public trust in the process and provide the new constitution with stronger democratic foundations. The preceding theoretical and empirical analysis suggests several desirable features in the design of a constitution-making process where a special convention must be used. In the absence of preexisting constitutional regulation, one option is to call an ad hoc constituent convention in which all or a majority of its members are citizens selected by some random selection method. A citizens’ assembly is appealing on both pragmatic and normative grounds because it may prevent the partisan manipulation of the reform process and enhance impartial deliberation. In practice, however, the little experience accumulated so far on citizen conventions suggests that in most political contexts these bodies are likely to be initially rejected or their decisions blocked ex post by established political parties.95 When members of the convention are to be elected on a partisan basis, constitutional and legal provisions should be devised for securing that the new constitution emerges from a politically plural and consensual decision-making process. This requires, in the first place, an agreement between executive and legislature to call a convention. It also contributes to enforce institutional checks on the process that citizens have the right to petition a popular vote on whether to call a convention. In no case, however, should the executive be allowed to submit this question as a referendum proposal without congressional consent. It is desirable that the regulation of the election, tasks, and powers of the convention be approved by qualified majority vote in the legislature. A highly proportional method of election should also be used to elect delegates and it is advisable that the convention be constrained to make final decisions by qualified majority. The convention should be free to decide on the content of the new constitution, but until the latter is enacted, the previous constitution should be observed and the institutions created under it maintain their full powers. A design along these lines may reduce the probability that a single political force is able to control the convention, increase the costs of transgressing the legal framework, and provide constitutional courts with some guidance in case they are required to decide on whether procedural rules were observed during the process. However, as this article has shown using evidence from Latin America, the realization of a legal, pluralistic, and consensual model of constitutional replacement by convention ultimately depends on the balance of power among the political forces that organized and controlled the process. When the stakes are high (as they usually are in these cases) and the distribution of political power is asymmetric, conventions are prone to violating their legal mandate and assuming legislative or other state functions in the name of the sovereign people. I would like to thank Jon Elster, Roberto Gargarella, Andrew Arato, Joshua Braver, Hélène Landemore, Hanna Lerner, Claudio López-Guerra, and two anonymous reviewers for their comments and suggestions on previous versions of this article. Footnotes 1 In an analysis of 411 episodes of constitutional replacement that took place around the world between 1789 and 2005, Tom Ginsburg, Zachary Elkins, and Justine Blount found that in most cases the approval body was either a constituent convention (103 instances) or a constituent legislature (178 instances). See Tom Ginsburg, Zachary Elkins, & Justine Blount, Does the Process of Constitution-Making Matter?, 5 Ann. Rev. L. & Soc. Sci. 201, 212–213 (2009). 2 Empirical data on constituent assemblies used in this section and other parts of this article have been collected by the author and belong to an ongoing research project that will be released in the future as the Comparative Constitution Making Database. This database covers all constitutions adopted in the world between 1900 and 2015 that have been implemented during years of free and fair elections for the largest portion of their lives. As of August 2017, it contains complete observations on 127 of 132 episodes that satisfy this definition. In 124 of the 127 observations a deliberative body was formed to approve the final text of the constitution. Information on the database can be obtained from the author upon request. 3 Some authors use the term “constitutional convention,” common in the American constitutional tradition, to denote constitution-making bodies commissioned to adopt a new constitution. See Jon Elster, Legislatures as Constituent Assemblies, inThe Least Examined Branch: The Role of Legislatures in the Constitutional State 182 (Richard W. Bauman & Tsvi Kahna eds., 2006). As Roger Hoar points out, however, the adjective “constitutional” is confusing because it may refer to both the specific task of the assembly and its status within the existing legal order. SeeRoger Hoar, Constitutional Conventions. Their Nature, Powers, and Limitations 30 (1917). As we will see, special conventions may or may not be authorized by the existing constitution. For this reason, throughout this text I use the generic term “constituent convention” or simply “convention” to refer only to the nature of the task of this body. 4 This includes as conventions assemblies whose central task was drafting a new constitution although they occasionally performed legislative activities for a short period of time until a new legislature was elected. In this regard, I follow Marc Kruman’s perspective in Marc W. Kruman, Between Authority and Liberty: State Constitution Making in Revolutionary America 15, 22–23 (1997). 5 A total of ninety-two of a sample of 124 constituent assemblies created between 1900 and 2015 were popularly elected, almost always by direct election. Of these, twenty-eight correspond to constituent conventions and sixty-four to constituent legislatures. Only eight of thirty-seven conventions were appointed and 1 used a mixed method of selection. Only two of the sixty-nine constituent legislatures were appointed and three used a mixed method of selection. Data is from the Comparative Constitution Making Datababase. See supra note 2. 6 Data from the Comparative Constitution Making Datababase. See supra note 2. 7 Hoar, supra note 3, at 4. 8 See Robert F. Williams, Are State Constitutional Conventions Things of the Past? The Increasing Role of the Constitutional Commission in State Constitutional Change, 1 Hofstra J. Pub. Pol’y 1 (1996). 9 John Dinan, The American State Constitutional Tradition 7, 8 (2006). 10 In a sample of 160 constitution-making episodes around the world from 1780 to 2012, Jonathan Wheatley and Fernando Mendez find that twelve of the twenty-three events that used constituent conventions were located in the Latin American region. SeeJonathan Wheatley & Fernando Mendez, Patterns of Constitutional Design: The Role of Citizens and Elites in Constitution Making 29 (2013). In my own database I also find that most conventions were created in the Latin American region, although in a lower proportion (fifteen out of the thirty-seven events that used this type of assembly). See supra note 2. 11 See Gabriel L. Negretto, Constitution Making and Constitutionalism in Latin America: The Role of Procedural Rules, inComparative Constitutional Law in Latin America 17 (Rosalind Dixon & Tom Ginsburg eds., 2017). 12 SeeWalter F. Dodd, The Revision and Amendment of State Constitutions (1921). 13 Data is from the Comparative Constitution Making Database. See supra note 2. 14 A constitution was considered to be new when its drafters claimed it was new and when state institutions and official sources in the country acknowledged the text as such. See Gabriel L. Negretto, Constitution Making in Comparative Perspective, inOxford Research Encyclopedia of Politics (2017). Democratic years were coded based on Carles Boix, Michael Miller, & Sebastian Rosato, A Complete Dataset of Political Regimes, 1800–2007, 46 Comp. Pol. Stud. 1523 (2010). 15 The 2010–2011 Icelandic constitutional assembly is the only case of a convention elected in a long-standing democracy outside Latin America for the purpose of replacing the existing constitution. So far, however, the constitution produced by this convention has not been put into force. 16 Data from the Comparative Constitution Making Datababase. See supra note 2. 17 Elster, supra note 3; Jon Elster, Securities Against Misrule: Juries, Assemblies, Elections 3 (2013) [hereinafter Elster, Securities Against Misrule]. 18 Elster, supra note 3, at 182. 19 SeeKruman, supra note 4, at 22. Hoar, supra note 3, at 4, also includes in this category the 1776 constitutions of North Carolina, Georgia, and New Jersey. 20 See Karlo Mirth, Croatia’s Independence 32(33) J. Croatian Stud. 5 (1992), and Advig Kiris, Restoration of the Independence of the Republic of Estonia: Selection of Legal Acts, Ministry of Foreign Affairs of the Republic of Estonia (1991). 21 On constitution-making in authoritarian regimes, see Gabriel L. Negretto, Authoritarian Constitution Making: The Role of the Military in Latin America, inConstitutions in Authoritarian Regimes (Tom Ginsburg & Alberto Simpser eds., 2014). 22 The enactment of Slovenia’s 1991 constitution could be used as an example of the former and the 1998 Albanian constitution of the latter. 23 See Hellen Krunke, Formal and Informal Methods of Constitutional Change in Denmark, inEngineering Constitutional Change: A Comparative Perspective on Europe, Canada and the USA 73 (Xenophon Contiades ed., 2013). 24 Roger D. Congleton, Improving Democracy Through Constitutional Reform. Some Swedish Lessons (2012). 25 See Markku Suski, Finland, inHow Constitutions Change. A Comparative Study 87 (Carlo Fusaro & Dawn Oliver eds., 2011). 26 The 1942 constitution was created from an initiative made by two-fifths of the whole legislature, the 1952 constitution from agreement between the main parliamentary parties, and the 1967 constitution from a text proposed by the Colorado and Nacional parties and passed by the legislature. Negretto, supra note 11. 27 See Giovanni Biaggini, Switzerland, inHow Constitutions Change. A Comparative Study 303 (Carlo Fusaro & Dawn Oliver eds., 2011). 28 For the historical and conceptual evolution of the constituent power theory, seeMartin Loughlin, The Idea of Public Law (2003), and Andreas Kalyvas, Popular Sovereignty, Democracy, and the Constituent Power, 12 Constellations 223 (2005). 29 Joel Colon-Rios, Weak Constitutionalism. Democratic Legitimacy and the Question of Constituent Power (2012). 30 See Thomas Jefferson, Notes on the State of Virginia, inWritings 124, 249–250 (Merrill D. Peterson ed., 1984). 31 SeeBruce Ackerman, The Future of the Liberal Revolution (1994). 32 Elster, supra note 3, at 185. 33 Elster, supra note 3, at 186–189. 34 This argument is derived from a more general position about procedural rules. According to Elster, these rules should be designed with an eye to removing obstacles to good decisions, without pursuing an ideal of what those decisions should be. Elster, Securities Against Misrule, supra note 17. 35 SeeElster, Securities Against Misrule, supra note 17, at 212. 36 See Elster, supra note 3, and Elster, Securities Against Misrule, supra note 17. Also see Jon Elster, Forces and Mechanisms in Constitution-Making, 45 Duke L. Rev. 364 (1995) [hereinafter Elster, Forces and Mechanisms], and Jon Elster, Clearing and Strengthening the Channels of Constitution-Making, inComparative Constitutional Design 15 (Tom Ginsburg ed., 2012) [hereinafter Elster, Clearing and Strengthening]. 37 So far, assemblies entirely composed of randomly selected citizens have been used only in electoral, not in constitutional, reform. The 2012–2014 Irish convention, responsible for proposing amendments, had a mixed composition: two-thirds of its members were randomly selected citizens and one-third appointed politicians. The 2010 Icelandic convention is the only example of a convention elected on a non-partisan basis. 38 As already mentioned, ninety-two of a sample of 124 constitution-making bodies have been popularly elected between 1900 and 2015 (see supra note 5). Of these, fifty-nine (twenty of which correspond to conventions, thirty-seven to constituent legislatures, and two to other bodies) were elected by PR formulas. Data is from the Comparative Constitution Making Database. See supra note 2. 39 Specifically, only thirty-two of a sample of 124 constitution-making bodies created between 1900 and 2015 have used a qualified majority rule to pass the constitution. Among these cases, however, twenty-seven correspond to constituent legislatures and only five to conventions. Data is from the Comparative Constitution Making Database. See supra note 2. 40 See Ulrich Preuss, The Exercise of Constituent Power in Central and Eastern Europe, inThe Paradox of Constitutionalism: Constituent Power and Constitutional Form 143, 218–219 (Martin Loughlin & Neil Walker eds., 2007). 41 Something similar happened with the 1998 Albanian constitution, whose draft was made by a representative constitutional commission rather than by the parliament as a whole. See Scott Carlson, The Drafting Process for the 1998 Albanian Constitution, inFraming the State in Times of Transition: Case Studies in Constitution Making 311 (Laurel E. Miller ed., 2010). 42 SeeGabriel Negretto, Making Constitutions: Presidents, Parties and Institutional Choice in Latin America (2013). 43 Prohibiting constitution-makers from competing in future elections may reduce the influence of their personal interests in constitutional choice but is not likely to prevent them from advancing the interests of their parties. See Elster, Forces and Mechanisms supra note 36, and Jon Elster, Ways of Constitution-Making, inDemocracy’s Victory and Crisis 123 (Axel Hadenius ed., 1997) [hereinafter Elster, Ways of Constitution-Making]. 44 On the longer duration of detailed and policy-oriented constitutions, see Christopher W. Hammons, Was James Madison Wrong? Rethinking the American Preference for Short, Framework-Oriented Constitutions 93(4) Am. Pol. Sci. Rev. 837 (1999), and Ginsburg et al., supra note 1. 45 See Negretto, supra note 11. 46 Ginsburg et al., supra note 1, at 213 47 SeeElster, Securities Against Misrule, supra note 17. 48 See Negretto, supra note 11. 49 See William Parlett, The Dangers of Popular Constitution-making 8 Brook. J. Int’l L. 193 (2012); David Landau, Constitution Making Gone Wrong, 64 Ala. L. Rev. 923 (2013); and Gabriel L. Negretto, Constitution Making in Democratic Constitutional Orders: The Challenge of Citizen Participation, inLet the People Rule? Direct Democracy in the Twenty-First Century 21 (Saskia Ruth, Yanina Welp, & Laurence Whitehead eds., 2016). 50 SeeJohn Alexander Jameson, A Treatise on Constitutional Conventions; Their History, Powers and Modes of Proceeding 10 (1887), and Hoar, supra note 3, at 30. 51 SeeJameson, supra note 50, at 209–269; and Hoar, supra note 3, at 30–37 and 38–57. 52 Id. 53 See the Pennsylvania Supreme Court ruling Wells v. Bain (1872), cited in Hoar, supra note 3, at 16–17. 54 SeeJameson, supra note 50, at 218–237; Dodd, supra note 12; and Hoar, supra note 3, at 20–22. 55 See Paul Thompson, Is There Anything Legal About Extralegal Action? The Debate over Dorr’s Rebellion, 36 New Eng. L. Rev. 385 (2001). 56 See Luther v. Borden (1849), cited in Jameson, supra note 50, at 224–242. 57 Whereas the 1789 Pennsylvania Convention was called after the legislature made an informal inquiry to determine whether the people wanted a convention, the 1791 Delaware Convention was called by the legislature and elected by the people without any initial authorization. SeeJameson, supra note 50, at 213–216. 58 Jameson, supra note 50, at 216–218. 59 SeeHoar, supra note 3, at 51–57. 60 See different views of Dodd, supra note 12, ch. 3, and Hoar, supra note 3, ch. 5, in analyzing the precedents about who has the authority to call and bind conventions. 61 SeeDodd, supra note 12, at 68–70. 62 Jameson, supra note 50, at 490–497. 63 Hoar shows that in several cases where the convention engaged in ordinary lawmaking, state court decisions invalidated those acts not because the convention had no power to do so but because it did not submit the decision to popular approval. SeeHoar, supra note 3, at 144. 64 SeeJames Madison, Alexander Hamilton, & John Jay, The Federalist Papers (Penguin Edition) 264 (1987). The expression “the people themselves” is used by Madison in The Federalist No. 40 in spite of the fact that the ratification process was indirect, through the election of delegates to state conventions. 65 See Richard Kay, The Illegality of the Constitution, 4 Const. Comment. 57 (1987); Akhill Amar, The Consent of the Governed: Constitutional Amendment Outside Article V, 94 Colum. L. Rev. 457 (1994); Bruce Ackerman & Neal Katyal, Our Unconventional Founding, 62(2) U. Chi. L. Rev. 478 (1995). 66 For a recent restatement of this argument, seeAndrew Arato, Post-Sovereign Constitution Making (2015). 67 SeeMadison et al., supra note 64, at 313–314; James Madison, Letter to Thomas Jefferson, February 4, 1790, inWritings 473, 474–477 (1999). 68 See John R. Vile, American Views of the Constitutional Amending Process: An Intellectual History of Article V, 35 Am. J. Legal Hist. 44 (1991); Walter E. Dellinger III, The Recurring Question of the “Limited” Constitutional Convention, 88 Yale L.J. 1623 (1979). 69 SeeDodd, supra note 12, at 188. 70 By contrast, the two Latin American cases in which constitutions were adopted by constituent legislatures (Uruguay in 1952 and 1967) show no variation in terms of the relevant factors analyzed in this article. They came into existence by means of a constitutional authorization and did not transgress their mandate. 71 In previous works, I considered the 1994 constitutional reform in Argentina as another instance of recent democratic constitutional replacement in Latin America. A reconsideration of the evidence, however, has led me to conclude that it should be coded as a constitutional amendment following the definition provided supra note 14. 72 This procedure was amended in 2002, but only to include a popular referendum to ratify amendments. 73 See Carlos Böhrt Irahola, El Proceso Constituteyente Boliviano, inLos Procesos Constituyentes Boliviano y Ecuatoriano: Analisis Comparativo y Prospectiva [Bolivian and Ecuadorian Constitution-Making Processes: Comparative and Prospective Analysis] 3 (Carlos Böhrt Irahola & Norman Wray Reyes eds., 2013) (unpublished manuscript, on file with the author). 74 The president had the power under the existing constitution to call a referendum on matters of national importance, which clearly included submitting the question of whether the existing constitution should be replaced and whether a convention should be elected to perform this task. SeeNegretto, supra note 42. 75 SeeNegretto, supra note 42. 76 See Decree 1926 of August 24, 1990, available athttp://www.suin-juriscol.gov.co/viewDocument.asp?id=30014406. 77 See Ruling 138, available at http://www.suin-juriscol.gov.co/viewDocument.asp?id=30014406. 78 See Ana María Bejarano & Renata Segura, Asambleas Constituyentes y Democracia: Una Lectura Crítica del Nuevo Constitucionalismo en la Región Andina [Constituent Assemblies and Democracy: A Critical Reading of the New Constitutionalism in the Andean Region], 79 Colombia Internacional 19 (2013). 79 The possibility of constitutional replacement by constituent convention was incorporated in 2004 without fulfilling Article 230 of the 1967 constitution. SeeJorge Lazarte, La Asamblea Constituyente: Un Nuevo Comienzo [The Constituent Assembly: A New Beginning] 39 (2006). 80 SeeRoberto Viciano Pastor & Rubén Martínez Dalmau, Cambio Político y Proceso Constituyente en Venezuela [Political Change and Constituent Process in Venezuela] (1998–2000) (2001); Allan Brewer Carías, Golpe de Estado y Proceso Constituyente en Venezuela [Coup D’état and Constitution-Making Process in Venezuela] (2002) [hereinafter Brewer Carías, Golpe de Estado]; Allan Brewer Carías, Dismantling Democracy in Venezuela. The Chávez Authoritarian Experiment (2007) [hereinafter Brewer Carías, Dismantling Democracy]. 81 In September 2006 the government party managed to impose as the first rule of procedure that the assembly was the holder of national sovereignty. See also regulation of this issue in the enabling congressional law, available athttp://pdba.georgetown.edu/Electoral/Bolivia/Leyes/LeyConvocatoria.pdf. 82 See Fabrice Lehoucq, Bolivia’s Constitutional Breakdown, 19(2) J. Democracy 46 (2008); Landau, supra note 49. 83 There were also conflicts regarding the time frame imposed on the convention, which the latter transgressed. 84 See Böhrt Irahola, supra note 73. 85 SeeNegretto, supra note 42. 86 See Ruling 138, supra note 77. 87 SeeNegretto, supra note 42, at 186. 88 From July 15 to December 1, 1991, a thirty-six-member commission appointed by the assembly was in charge of exercising legislative functions until the members of the newly elected congress took office. 89 Somewhat contradictorily, however, in January 1999 the Court had already decided that calling the convention through a referendum outside the mechanisms of revision of the existing constitution was valid because the constituent power of the people could not be limited by amendment procedures meant to be observed by the constituted powers. SeeViciano Pastor & Martínez Dalmau, supra note 80. 90 See Ricardo Combellas, El Proceso Constituyente y la Constitución de 1999 [The Constituent Process and the 1999 Constitution], 30 Politeia 193 (2003). 91 See Allan Brewer-Carías, El inicio del proceso constituyente en Ecuador en 2007 y las lecciones de la experiencia Venezolana de 1999 [The Start of the 2007 Constituent Process in Ecuador and Lessons from the 1999 venezuelan experience], 14 IurisDictio 109 (2007). 92 In regard to the judiciary, the convention established that the Supreme Court, the Supreme Electoral Court, and the Constitutional Court would continue in their functions unless it decided otherwise. See Böhrt Irahola, supra note 73. 93 While still qualifying as an electoral democracy (with a score greater than six), Bolivia went from an average score of eight during the period 2004–2008 to a score of seven in 2009 on the Polity IV scale (2015). 94 Such is the case of Venezuela after 2006, and Ecuador after 2007, according to Polity IV (2015). 95 SeeAlan Renwick, After the Referendum: Options for a Constitutional Convention (2014). © The Author(s) 2018. Oxford University Press and New York University School of Law. All rights reserved. For permissions, please e-mail: firstname.lastname@example.org This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
International Journal of Constitutional Law – Oxford University Press
Published: May 12, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera