3.5.1 Approving and ratifying the constitution

The constitutional referendum is but one form of the referendum; others include referendums on secession or merger with an existing state, legislation, treaties, and policies. The focus here is on the constitutional referendum on fundamental changes to an existing constitution, or the adoption of a new constitution.

Two important elements in designing a constitution-making process are the institutions and rules for making decisions on the constitution. Often the choice is dictated by political and legal traditions. An obvious distinction is between unitary and federal states; the constitution of a federation affects more than the federal government, necessitating a more complex set of institutions and rules for voting. In many cases, the deliberative body that made the draft, usually the legislature or the constituent assembly, also adopts and ratifies it. Britain and some former colonies dominated by the notion of parliamentary sovereignty leave constitutional issues to the legislature. In civil law systems, there is a preference for a referendum (though there are a number of examples of parliament making the final decision). In states under some kind of international trusteeship or custody, there is likely to be a constituent assembly or a referendum (such as in Iraq [2006]), unless there have been recent general elections to the body making the constitution (as in Cambodia and Namibia).

The purpose of this section is to examine the necessity and desirability of a referendum for the adoption of the constitution. Rules for decision-making by constitutional commissions, legislatures, and constituent assemblies have been discussed in previous sections. As a rule, a referendum ratifies a decision made by another body, usually a political body, the legislature, or the constituent assembly. (An exception is the current Somali process, in which the draftconstitution is required to be submitted directly from a constitutional commission to the people under the Somalia Transitional Charter—though this is unlikely to be feasible in the current circumstances in Somalia.) In principle it is desirable that the draft constitution should be considered and approved by a more deliberative body prior to the referendum, since the referendum does not normally provide a proper opportunity for the canvassing of views on issues in the constitution, or the accommodation of different perspectives. And the only choice in the referendum is “Yes” or “No,” not a nuanced answer.

Uses of referendum

It is possible to use the referendum to resolve a particular controversy about the design of the process or a substantive issue. It has been used, particularly in Latin America, to ask the people if they would prefer a constituent assembly to draft the constitution, or to secure the mandate for negotiations on a new constitutional order (as in the referendum of the white community in South Africa), on the mandate of the constitution-making body (as in France in 1957), or on secession and independence (as in Saint Kitts and Nevis in 1998; Timor-Leste in 1999; Bougainville, Eritrea, Montenegro, New Caledonia, South Sudan in 2006; it is also a possibility for Canada, in relation to Quebec, as result of a Canadian Supreme Court decision on the subject). A similar referendum in Kurdistan in 2005 on independence was held by the regional authorities, to secure, as expected, a big “Yes” vote to strengthen its negotiating position on autonomy with respect to the rest of Iraq. The referendum to settle a contentious issue is best exemplified by the decision of the Special Majlis (constituent assembly) in the Maldives to adjourn its proceedings to refer the major contentious issue—the system of government—for resolution by the people. Both the Uganda [1995] and Kenya [2005] processes provided for similar possibilities, but these were not taken up, either to put pressure on the constitution- making body to reach consensus, or because they were daunted by the complexity of the issues. But Uganda used the referendum five years after the constitution to settle the outstanding question of whether the country should move to a multiparty system. Decisions on the retention or abolition of monarchy have been made by referendum in several countries (such as Greece, Italy, and colonial Rwanda).

Our principal concern, however, is with the referendum for a new constitution or major amendments, for at least two reasons. First, for the purposes of constitution-making, the referendum makes people a key institution, whose approval is necessary to bring the draft into force. This role of the people flows from the sovereignty vested in them, and is a manifestation of the fundamental component of the principle of self-determination. Second, for certain constitutional changes, the direct participation of the people in the referendum is seen as a necessary condition for the validity of constitutional change. Courts in India and Kenya have held that fundamental constitutional changes cannot be made by the ordinary process of constitutional amendments, especially when undertaken by the legislature. And, a fortiori, a constitution cannot be replaced by the ordinary amendment process. These defects can be perhaps be cured if an assembly has been elected democratically for the express purpose of amending fundamental features or of bringing in a new constitution. A referendum, it follows, would put the attempt beyond doubt.

Prevalence of referendum

Professor Markku Suksi, a foremost authority on the referendum, notes that more than half of the constitution-making processes from 1998 through 2007 used the referendum to decide on new constitutional text. But at the level of states (ignoring subnational entities), new constitutions, or constitutions that seem to be the result of a more specific constitution-making process rather than a regular amendment process, have been decided by means of the referendum only in some fourteen cases (Albania [1998], Sudan [1998], Venezuela [1999], Qatar [2003], Rwanda [2003], Cyprus [2004], Burundi [2005], Iraq [2005], Kenya [2005; 2010], Democratic Republic of the Congo [2006], Serbia [2006], Kyrgyzstan [2007], and Thailand [2007]). Suksi also estimates that at least half of the world’s constitutions require a referendum for constitutional change (Suksi 2008).

Status of referendum

Apart from the subject matter, referendums can be categorized by their status. A referendum may be mandatory, required by the constitution or some other law. In these cases the results of the referendum are usually binding. A referendum may be advisory, to enable the government or parliament to gauge public opinion. In countries such as Britain in which referendums are neither mandatory nor binding, there may nonetheless exist an unwritten convention that certain important constitutional changes will be put to a referendum and that the result will be respected. The Kenyan referendum [2005] was probably advisory (if positive, the actual change to the constitution would have required parliamentary approval), but the rejection effectively killed the constitution-making process.

Some countries have elaborate rules on who may call a referendum, and when; a referendum can have a constructive or a disruptive influence on politics, and can certainly be used in a partisan way to undermine regular political institutions and processes. Sometimes it is necessary to get the permission of the courts to hold a referendum (as in Hungary). A court in Zimbabwe has held that the president can call a referendum at any time (and submit any draft constitution). Since only the legislature has any constitutional amendment power, the referendum can only be advisory. (As it happens, the Zimbabwe draft was rejected and the president made no attempt to take it or an amended version to the legislature.) For the purposes of constitutional amendment or replacement, the rules are normally relatively straightforward, and compulsory.

Rules regarding voting

How are the results of the referendum computed? On this critical matter, there is no single answer. In most systems a simple majority of those voting suffices (but in the Republic of the Congo, an absolute majority of those voting is required, and in Lithuania some provisions, touching on the character of the state, require a three-quarters majority). Some laws prescribe that a minimum percentage of registered voters must have voted, and many referendums have been lost even though the majority of those who voted said “Yes,” because of low turnout. (Ireland and Italy are among the countries that require a minimum turnout.) A few, particularly federations, require that apart from an overall majority, there must be a significant spread of support throughout the country. This rule was applied in the referendum in Kenya [2010], requiring an overall majority of those voting and 25 percent support in five out of eight provinces. (Kenyan provinces have ethnic dimensions.) In the Iraqi referendum of 2005, the requirement was both a majority of “Yes” votes nationwide and that not more than two governorates (out of eighteen) have a “No” vote by two-thirds or more of the registered voters. (This figure was chosen rather than the original and less onerous requirement of two-thirds of the actual voters, the change giving a veto to small, but compact, groups—a rule originally meant to help Kurds in their negotiations for autonomy, but which nearly enabled the Sunnis to torpedo the constitution.)

Debating referendums

With the above framework for referendums, the question is what role should there be for the referendum? We need to ask questions about such issues as: how important is it to give the final word on constitutional issues to the people; about the place of the referendum in what is often a long and complex process covering a variety of issues, some controversial; the nature and fairness of the referendum campaign and process; the fairness of the referendum; and its impact on politics and nation-building. Is the referendum about the constitution, or is it really about other issues? Are there particular circumstances when a referendum might be particularly valuable—and others when it is unnecessary, even counterproductive? In order to answer these and other pertinent questions, we look at the advantages and disadvantages that are commonly advanced about referendums.

Arguments in favor of referendum

People’s sovereignty and democracy; conformity with international norms—self- determination

The preamble of most constitutions says, “We the People give ourselves this constitution,” or something to that effect. Modern democratic theory and emerging international norms, such as self-determination, proclaim that the sovereignty of a nation is vested in the people. Since the constitution is the supreme law, a manifestation of national sovereignty, it is appropriate that the final word should rest with people.

People’s participation

Closely connected with the above point is the right of the people to participate in the constitution- making process. (See part 2.2.1.) Traditionally, the entire scope of people’s participation was the referendum, generally after the constitution was made in considerable secrecy, by a few men drawn from the upper classes. Today, as we know, people participate in many ways, but often they play a small role in decision-making—except when there is a referendum.

Check on constitution-makers

The final word is often a commentary on, or an assessment of, the results of preceding efforts. If the process has been participatory, people have a chance during the referendum to decide whether their views have been sufficiently taken into account. There are several examples of people rejecting the constitution: Albania [1998], Zimbabwe [2000], Kenya [2005]. Sometimes the rejection kills efforts at reform, but more often it leads to a better process, as in Albania and Kenya (although after a lapse of some years).

Accountability of constitution-makers

The referendum can be a mechanism of accountability. It provides an opportunity for wide- ranging debate, and if the sponsors of the draft, usually members of the government, lose, they are expected to resign. In practice this has not happened. In Kenya [2005] the president dismissed the “No” faction of the cabinet; in Zimbabwe the president intensified his oppression of the people; in Albania the government stayed on and rigged the next election. But it is likely that the regime loses some moral and political authority.

Incentives for compromise to ensure general support of all ethnic and other communities

It is possible that if the process includes a referendum, this leads to caution on part of constitution-makers involved in earlier stages of the process, making them more sensitive to views of the people and more inclined to resolve differences among themselves. It is possible that in South Africa [1996] the provision for a referendum if the constituent assembly was not able to agree (by a two-thirds vote) might have persuaded the National Party to make more concessions to the African National Congress than it might have done otherwise, and equally that the African National Congress might not have wanted the agreement with the National Party, negotiated over a long period of time, to collapse through extremist influence on the referendum. However, it does not seem that the Kenyan politicians, acting through the parliamentary select committee, were particularly concerned about the reaction of the people (even shortly before the 2010 referendum was due), but the committee of experts, realizing that the last word lay with the people, felt emboldened to restore some people-oriented provisions that had been removed by the committee.

Change after careful consideration

Insofar as the referendum provides another opportunity to reflect on the draft constitution, and involves people who might not previously have been engaged in the process, it may prevent hasty, ill-considered, or opportunistic alterations to the constitution.

People’s knowledge of constitutional matters

The referendum campaign is an intense period of concentration on political and constitutional issues, explanations of the draft constitution, and arguments about its pros and cons. This provides an incentive as well as a wonderful opportunity for the people to learn about constitutional values and institutions, and perhaps their own responsibilities as citizens.


Professor Suksi remarks that several recent referendums have taken place in countries in, or coming out of, conflict. For various reasons, he says, political or military leaders in these countries use referendums to secure the broadest possible legitimacy for a new constitutional arrangement and the greatest possible visibility for the constitutional solution. The referendum may, under such circumstances, be the final resort, either due to opportunism on the part of the leaders of the state to legitimize their position or because of an honest attempt to re-create public authority by a reference to the root of legitimacy, the people (Suksi 2008).

Legitimacy—people’s aspirations: Putting the matter beyond doubt

Underlying many of the above arguments is the legitimacy that is supposed to flow from people’s endorsement of the constitution. Words such as “ownership” and phrases such as “the commitment to defend the constitution” are used frequently. There is no doubt that people who have been given, and have taken, the opportunity to vote on the constitution will feel a greater stake than they would in one bestowed on them. The sponsors and supporters of the constitution often point to the enthusiasm of the people, emphasizing the significant majority for it, while its opponents feel hesitant to criticize it.

However, there is little empirical evidence that the euphoria or the hesitation lasts for long— unless the constitution delivers—or that people would take to the streets to defend the constitution. Thailand’s 1997 constitution resulting from a highly participatory process was overthrown easily enough. Even more striking was the lack of any popular resistance to the refusal of the Eritrean president to bring the 1995 constitution, which had been crafted with full public participation, into force. On the contrary, there is the risk that if the high expectations are not met soon, the support for the constitution will disappear. The next section discusses arguments against the referendum.

Arguments against referendum

A constitution is not a one-issue matter, but a complex set of values and institutions

A common worry about the referendum is that the voters will not understand the constitution or appreciate the significance of contentious issues. They may, indeed, reject the constitution on the basis of one or two issues that are peripheral to it. It is safe to say that in few cases have the vast majority of the people understood the real significance of the constitution, or the compromises that have gone into creating it.

Danger of misleading campaigns

Most referendum campaigns are dominated, if not hijacked, by politicians. Experience shows that many of them do not understand the constitution, and in many cases have not even read it. Ignoring the central objectives and themes of the constitution, they blow a minor provision out of all proportion, particularly if it appeals to the emotions, and they make the referendum a one- issue matter. They are not above deliberately misleading the people; the campaign becomes a tissue of lies—the antithesis of deliberative democracy. If the politicians are divided about the constitution, their supporters will follow them and make little effort to understand the true issues.

Voting for the wrong reasons

The party politicization of the campaign can infect the people, so that the motives for their votes have little to do with the content of the constitution (or even misunderstandings about it). There are, as it were, purely political party considerations that can delegitimize the sponsors, usually the government. In countries where parties are unstable and fluid, the considerations are ethnic. The 2005 and 2010 referendums in Kenya were marked by extreme appeals to ethnicity; their effectiveness was reflected in the voting patterns. The general feeling was that the campaigns were really about the 2007 and 2012 general elections. The referendums became politics by another name.

Dangers of manipulation and intimidation, and seeking spurious legitimacy

In many countries, especially in developing areas, the referendum campaigns and voting have many of the hallmarks of elections—use of official resources for campaigns; intimidation and the use of the militia; bribery and other forms of electoral corruption, including the stuffing of ballot boxes. In Rwanda the voters were deliberately given only two weeks for debate, attempts at independent assessment were suppressed, and the fear of another genocide was evoked by the government if the constitution were rejected. The situation was worse in Sudan, where the draft, as presented by the commission, was altered by the president and then submitted to a highly manipulated and controlled referendum. Recently a similar charade was enacted by the military authorities in Myanmar (formerly Burma). In none of these states was the process open or participatory; instead it was marked by censorship and coercion.

Deeply divisive or polarizing effect of referendum in multiethnic states

The points above suggest that, far from the referendum bringing people around a constitution and promoting national unity, it can be deeply divisive. The situation after the referendum can become worse than before. (This is illustrated by the experiences in Cyprus, Iraq, and Kenya.) These experiences also showed the majoritarian bias of the referendum.

A better form of public participation is at earlier stages of the process: People as decision-makers

If an objective is the participation of the people, it is more effective to bring the people into the process at earlier stages, when there are discussions about the reform agenda and preliminary decisions on the constitution are being made. This will engage them more deeply and make them the real decision-makers.

Referendum can be used to defeat the results of a fair and participatory process

The referendum is sometimes used to defeat the results of a fair and participatory process. In Sudan [1998], Zimbabwe [2000], and Kenya [2005], key government officials not in agreement with the drafts produced through predetermined procedures put to a referendum their own versions of the draft. Although in some cases the doctored drafts were rejected, the original drafts were not enacted. In Kenya it took several years, and another lengthy process, to install the original draft, and even then it was not promulgated in its entirety.

In multiethnic states, the better solution is to negotiate in good faith

In multiethnic states, the strongest argument against a referendum is that a laboriously and carefully negotiated settlement between different communities can come unstuck. These settlements are almost always controversial, yet they are the only basis on which agreement is possible. But disgruntled members of a community, particularly of the majority community, can use the referendum campaign to mobilize opposition with the use of racist propaganda. In a country with a troubled ethnic history, such appeals are likely to succeed. The result is deep resentment in one or more communities and the general worsening of ethnic relations. Equally likely is that some leaders would not even agree to a settlement, fearing that they would be accused by their communities of selling out during the referendum campaigns. The danger of referendum politics sharpening ethnic divisions is real. The recent Canadian experience in getting a new constitutional order shows the incompatibility of a negotiated compromise with the outcome of referendum.

Legitimacy comes from the fairness and effectiveness of the constitution

Some of the most successful constitutions have not had the benefit of a referendum, such as those of Canada, Germany, Hungary, India, Japan, Mauritius, and South Africa, while many constitutions endorsed by referendum have performed poorly or been replaced (e.g., those in Eritrea, Sudan, and Thailand). Fairness depends on the context; in the case of a multiethnic state it means that there is respect for all communities, and individual as well as community rights are protected within an overarching national identity. Effectiveness means that the status of the constitution as supreme law is honored and therefore the rule of law prevails.

Importance of the question

It is well understood that, deliberately or by carelessness, how the question is put can affect the result. When Australians were asked about the abolition of the monarchy, it is probable that a majority favored this step, but the question as framed faced them with a choice about how to replace the British monarch as head of state. This issue divided the republican vote. In Kenya [2010] the question “Do you approve of the proposed constitution?” was arguably less clear than one showing that the choice was between the existing constitution and the proposed one.

The United Kingdom electoral commission criticized a referendum on powers for the Welsh assembly for being ambiguous and using phrases (such as “devolved powers”) that people did not understand.

Comment on the debate

The arguments for and against the referendum show that while in theory it is democratic and basic to the sovereignty of the people, it poses serious practical problems. It may well be necessary in some contexts and unwise in others. It would be useful to make a careful analysis of the desired or expected objectives of the referendum before embarking on it, and to structure it accordingly. Meanwhile there are some obvious procedural improvements that should be instituted, particularly to make it fairer, more effective, and more educational.

Referendum reforms

  • One complaint about the referendum is that it gives limited choice, expressed as “Yes” or “No.” Is it possible to offer more choices? Options that may be considered are:

    • choices on a series of contentious issues (for example on abortion or on the electoral system) before the draft is finalized; or
    • choices between two drafts (for example, one based on the parliamentary cabinet system and the other on the presidential).
  • The draft constitution should be published well before the referendum date so that there is sufficient time for proper dissemination and discussion.
  • The question on which the voters have to decide should be clearly stated, and announced well in advance; testing questions ahead of time to see whether they are understood would be wise.
  • To avoid the hijacking of the campaign by politicians, an independent commission of eminent and respected citizens should be established to assist with dissemination of the draft, and the organization and funding of campaign activities, including:

    • promoting understanding of the purposes and procedures of the referendum, in cooperation with the electoral commission and other relevant bodies;
    • establishing or accrediting national “Yes” and “No” teams, drawn from academics and intellectuals, retired civil servants, professionals, and civil society, with appropriate gender, ethnic, and regional balance;
    • facilitating town hall-type meetings and media programs jointly with “Yes” and “No” teams to promote fair and open debate;
    • preparation, publication, and dissemination of materials to facilitate understanding of the draft, including for people with reading impairments, and in different languages where appropriate; and
    • engaging with monitors overseeing the campaigns and the voting.
  • A regulatory system for campaigns and for the conduct of voting, with a code binding on all political parties and other groups active in the campaign, should be established to ensure a fair campaign, including:

    • no hate speech;
    • no bribery of voters;
    • no deliberate misleading of voters and others about the draft;
    • no intimidation or use of force against those with opposing views or voters; and
    • no use of government resources other than through the referendum commission.
  • To make the referendum meaningful, the design of the constitution-making process should provide for the participation of the people in earlier stages of the process so that they become aware of the issues and the debates and are able to make an informed choice at the referendum.

Part 3: Institutions, groups, and procedures

The decision to leave the discussion of institutions and procedures for constitution-making processes to a late stage in this handbook was motivated by a conviction that the design of institutions ought to be determined in large part by their tasks. However, the process is often designed the other way round: the decision is expressed as something like “We must have a constituent assembly.” Other than recognizing that the general aim is to make a constitution, what the constituent assembly will do is spelled out only later.

It is also true that, in many countries, national traditions dictate what bodies will have responsibility for the process; indeed, for some readers this part may be of limited interest, because in their countries either custom or existing law is clear on the issue of institutions and processes. However, we venture to suggest that even for those readers this part may offer some insight: for example, not all constituent assemblies are made up or operate in the same way. There are even some differences in conducting referendums, and examining the experience of another country might be useful.

Increasingly, the constitution-making process is negotiated or designed, either in international conferences (Cambodia [1993] or Afghanistan [2004]), as part of interim arrangements (South Africa [1996] or Iraq [2005], Nepal [ongoing process]), in national conferences (Kenya [2005], francophone African countries), or through “roundtable” processes (in some Eastern European countries). In recent years there have been several comparative studies of constitution-making processes. This scholarship is aimed partly—even primarily—at drawing lessons about the most effective institutions and methods. The growing internationalization of processes means that decision-makers have access to several alternatives to their traditional approaches, and conscious decisions on the design are increasingly being made. Recent years have also seen new features of constitution-making processes, such as participation of both the international community and local citizens, which require new forms of institutions and procedures. Much can be learned from the successes and failures of countries that have pioneered innovations. The South African process, generally deemed to be successful, has been studied by several countries.

Having explored what is involved in making a constitution, we now turn to institutions and procedures. Systematic analysis of these also presents difficulties. Table 2 indicates that the same tasks may be performed by a range of different bodies. And drawing lines between different bodies is not easy. A constituent assembly is different from a parliament. But some parliaments carry out the same functions as the assemblies, and some constituent assemblies serve also as parliaments. Some constituent assemblies only debate a draft constitution prepared by someone else. Some carry out civic education, consult the public, prepare a draft, and then formally debate it. A commission may be quite small and expert, or rather large and not so expert, or even small and not expert! It may or not be independent—of government or of politics.

It is possible to view the United States constitutional convention in Philadelphia in the late eighteenth century as an early example of a constitutional commission (though we tend to classify it as a sort of constitutional assembly). Its mandate was to review and recommend changes to the confederal constitution, which would have been debated and possibly adopted under the mechanism for amendment of that constitution. (As is well known, the convention exceeded its mandate and established a new procedure for the adoption of its draft.)

We should warn against assuming that the same name indicates the same sort of animal in all countries, or that a different name indicates something different. A constitutional convention, a constitutional assembly, and a constituent assembly may be quite similar.

Our concern in this part is with how these bodies are formed, including their legal basis, how they operate—and how they might operate better.

Some principles of organization are essential. We have chosen again to focus on function rather than on form. We have grouped together institutions that generally perform rather similar functions. However, they also do tend to cluster in terms of size. So we begin with the big bodies that often carry out a lot of functions. They may indeed be the only constitution-making bodies in a given process: constituent assemblies, parliaments, and national conferences.

Then we turn to bodies with the core function of preparing constitutional proposals—including an actual draft constitution. They may do other things as well, such as civic education. These bodies might be called commissions, committees, or roundtables, and they might be (and usually are) specially formed, but might be longer-term bodies.

Next we look at administrative bodies. We discussed in part 2 the processes of administering and managing a constitution-making body. Here we look briefly at the institutions that carry out this role. We use the term “administrative management body” for these, although this is not the name any actual body is likely to have. These bodies will be called “secretariats or management committees” or something similar. Although the role this body plays is critical to the success of a process, it has largely been ignored in constitutional scholarship.

A lot of other official bodies may play limited roles in constitution-making. By no means will every constitution-making process have them, and many of these bodies will not have been planned from the beginning. They are a slightly “miscellaneous” collection, having little in common other than that they have some function in connection with the constitution. They include government departments, bodies that mange elections, bodies of experts, and the courts.

Many constitution-making processes require a referendum for the final decision on the draft, and less commonly while an effort is made to resolve differences on specific issues.

Box 35. Commissions or committees?

There is no real difference between these two words, at least in English. But usage may vary. In English it would be unusual to use “commission” to refer to a section of a larger body. English would refer to a “committee” of a legislature or assembly; Spanish might use “comisión.” “Commission” in English usage suggests a slightly more weighty body. We use “commission” for a separate body and “committee” for a part of a larger body.

How the bits may fit together

The case studies in appendix A suggest how these elements of a constitution-making process may fit together. No process will have every type of body or institution mentioned. Some may have only one official body. Here are some possible combinations:

  • Constituent assembly or legislature only
  • Commission → constituent assembly
  • Commission → constituent assembly → referendum
  • Commission → constituent assembly → legislature → referendum
  • National conference → legislature
  • National conference → referendum

Table 7: Comparison of a constitutional assembly and a parliament

Parliament (legislature)

Constitutional assembly

It functions by virtue of the constitution and according to the constitution

It makes the constitution—though this may sometimes be by virtue of law made under the authority of the existing constitution

It makes laws

It will not usually make ordinary laws, but it will make the constitution

It votes money for government (budget)

It has no control over finances

It holds the government accountable

It does not hold the government accountable

It is a body that represents the people—usually through elections and political parties—and functions as their representative, rarely with much active public involvement

It usually represents the people—but its members may be chosen other than through elections, or parties; it may involve the public in its activities far more than the legislature would; it is a manifestation of the sovereignty of the people

In a parliamentary system it constitutes a government (the government will usually be drawn from among its members)

It will not affect the formation of a government

Its own accountability to the people depends largely on periodic elections

It is unlikely to have any mechanism by which it is accountable to the people (other than in that decisions may be submitted to them through a referendum)

In any of these (and other possible) chains of institutions, other bodies are likely to be involved. Government departments and electoral bodies are the most obvious; special committees may be involved, as will (almost certainly) local and perhaps foreign experts. Civil society will offer input and perform a variety of roles.

Different patterns

In the contemporary period, the commission has been used extensively in Africa, including Tanzania [1965], Ethiopia [1994], Uganda [1995], Eritrea [1997], Zimbabwe [1999], Rwanda [2003], Kenya [2005; 2010], and Zambia (four times). We suggest why this may be so in part 3.2.3.

It has also been used in Afghanistan [2004], and in Pacific island countries such as Papua New Guinea [1975], Fiji [1997 and earlier], Bougainville [2004], and Nauru [2010]. States or territories associated with the United States tend to use constitutional conventions, though these may be preceded by commissions—as in American Samoa in 2010.

Asia, Europe, and Latin America have more often used only a constituent assembly. India led the way in Asia [1950] and was emulated by Nepal [1951 and ongoing process]; Indonesia had an abortive constituent assembly [1959]. France [1789] had perhaps the first true constituent assembly, and many European countries have followed suit, including Norway in 1814 and Italy after World War II, and some Eastern European countries in recent years. In the modern era, many European countries have carried out constitutional reviews through ordinary legislatures.

The USSR had a short-lived constituent assembly after the revolution. In Latin America the constituent assembly, without any preparatory commission, remains the major body.

Some francophone African countries, and a few others, have used the device of a national conference. The origin of this can also be traced back to the French constituent assembly— which again shows the difficulty of exact terminology.

3.1.1 Introduction

The bodies discussed here are mostly fairly large, and they are often charged with a number of functions: educating the public, collecting views, developing guiding principles, and drafting and adopting a constitution. (Some perform only a few of these roles.)

The distinctions among the various bodies are also not precise. A national legislature has a clear role. But a constituent assembly may also be a legislature. And the boundary between constituent assemblies and national conferences also is not a clear one. In this section we consider constituent assemblies and national legislatures together. National conferences, a phenomenon largely found in francophone Africa, are addressed separately.

3.1.2 Constitutional assemblies

Here we discuss bodies that fit this following broad description: a body designed to represent the nation, assigned—at a minimum—the task of debating in detail a draft constitution of the country, and of approving that draft. It may or may not also have the task of preparing the firstdraft, or have the final responsibility for passing it into law. We use the phrase “constitutional assembly” here because it is wider than the phrase “constituent assembly.” We include “regular” legislatures that are responsible for a new constitution or constitutional revision. If a comment applies specifically to legislatures, we make that clear. Comments that relate to “constitutional assemblies” apply to any representative body charged with making or revising a constitution.

The significance of the common phrase “constituent assembly” is that it refers to a body representing the people that is vested solely (or mainly) with “constituent power.” A Nigerian constitutional authority, B. O. Nwabueze, wrote “[Constituent power] is a power to constitute a frame of Government for a Community, and a Constitution is the means by which this is done. It is a primordial power, the ultimate mark of a people’s sovereignty” (Nwabueze 1974: 292). In some countries it is assumed that the people’s constituent power can be exercised by parliament; in some the tradition is that a separate body is required, and usually such a body does not have the power to make ordinary laws; the parliament alone retains that role. But views may change—especially because of the highly political atmosphere that often surrounds constitution-making. A Kenyan court in 2004 decided that a new constitution could be made only by a directly elected constituent assembly, or possibly a parliament elected for the purpose; failing this, a referendum was required—which was a highly questionable decision.

We can compare the main characteristics of a constitutional assembly and a parliament:

There is no single model of constitutional assembly, and it is not really possible to say that certain minimum requirements define such an assembly. Even names vary; many combinations of the words “constituent” or “constitutional” with such words as “assembly,” “convention,” “congress,” or even “conference” have been used to name (and describe) bodies whose primary responsibility is to change or make the constitution. In some languages there is no distinction between “constituent” and “constitutional.” But in languages that do have both terms, “constituent assembly” has a particular appeal because it implies that it is the people’s representatives who make and adopt the constitution. In Germany this phrase was rejected and “parliamentary council” was used—one of several measures designed to make it clear that it was not the final constitution for a united Germany that was being prepared.

Constitutional assemblies differ in size and composition and in how their members are chosen. They also vary in their roles, although they must at least discuss and adopt a constitution. Some are national bodies preparing a national constitution, while others are subnational bodies—such as the body that prepared the Bougainville constitution [2004], the Kashmir constituent assembly of [1956], and the assemblies of the individual states of the United States.

Legislatures and assemblies: The relationship

Possible relationships among special constitution-making bodies and “ordinary” legislatures are these:

  • There is no “regular” legislature at all, and all efforts are focused on the constitution-making process through the constitutional assembly.
  • The regularly elected legislature alone has the task of constitution-making or revision (and it is not elected specifically for that purpose).
  • The legislature is chosen by elections in which the constitution is the only or a major issue, and it has the task of making the constitution.
  • The legislature is also the “constituent assembly,” though it is analytically two separate bodies (e.g., as in India, Nepal, Papua New Guinea, and South Africa).
  • All the members of the legislature are members of the constitutional assembly but there are also other members of the constitutional assembly, and the two bodies sit separately; the regular legislature as such has no role in connection with the constitution.
  • All the members of the legislature are members of the constitutional assembly, but there are also other members of that assembly, and the two bodies sit separately. The regular legislature also has a role in connection with the constitution (as in Kenya [2005]).
  • There are two completely separate bodies, with no (or no significant) overlap in membership, and the regular legislature has no role in connection with the constitution (e.g., Bolivia and Uganda).
  • There are two completely separate bodies, with no (or no significant) overlap in membership, and the regular legislature does have a role in connection with the constitution.

If there is both a legislature and a constitutional assembly, each of which has a role in the constitutional process, how are these roles divided? The role of the legislature may be limited to forming the assembly and other organs of constitutional review. But the legislature may insist on having the last word (or a later word). So the constitution adopted by the assembly may have to go to the legislature for further approval (as happened in Kenya in [2005]). This is an unsatisfactory arrangement: if the assembly is inclusive, it is odd that its product should then go to the less-inclusive legislature. And if, as in Kenya, the members of parliament are also members of the assembly, they may be tempted not to engage in that body because they will have the chance to decide on the constitution later, in the legislature.

If one body operates as both legislature and constitutional assembly, how is the difference marked, if at all? In Ceylon (now Sri Lanka) the constituent assembly sat without a mace, the symbol of royal (governmental) authority. The two bodies could sit in separate places, but this would create logistical problems if both needed to sit on the same day. A different person might preside. The session for each might be marked by its own formalities—separate prayers, for example. The rules of procedure might be slightly different—but it would be hard to make them significantly different for similar types of activity. There is a good reason for having more demanding quorum rules for the constitution-making body.

In Nepal [ongoing process] the interim constitution provided for a committee of the constituent assembly to carry out the legislative work. The motive was to prevent the work of the constituent assembly being held up because the parliament needed to meet. However, this was never brought into effect. It seems there was an unwillingness to leave the parliamentary work to a committee. There were twenty-five parties in the assembly and all would want to be in the parliament, yet some had only one member.

Box 36. The nondeliberative constitution-making role of parliament

Parliament may have—or give itself—functions in relation to the constitution-making process other than debating the contents of the constitution. It may well have passed the laws setting up the process. It may have a role in appointing members of a constitutional commission. It may have to vote the necessary resources for the process. It may insist on overseeing the process, even if it cannot control it. In Kenya from 2000 to 2004 and again from 2008 to 2010 the legislature established a select committee that carried out these roles. The committee was important if amendments to the governing legislation had to be passed. Its makeup reflected the party composition of the parliament. In 2008–2010 the committee was also given specific roles by law (including decision-making roles about the content). The Somali process that began in 2009 emulated this model, but the role of its parliamentary committee was unclear.

Arguments for and against separating the legislature and the constitutional assembly

A constituent assembly may be seen as representing the people’s sovereignty. It may also be seen as a way precisely of ending legal continuity, marking a break with an autocratic past, or having a genuinely “home-grown,” or autochthonous, constitution rather than one that owes its legitimacy to its colonial history (as in Papua New Guinea [1975]). It has been argued that it is better to have a constituent assembly, which is somewhat different from the ordinary legislature, make the constitution, precisely because it is not the ordinary lawmaking body and does not have the same vested interest in the document that is to be drawn up; it should not “act as judge in its own cause.” This benefit will disappear if the constituent assembly turns itself into the legislature under the new constitution (as in Timor-Leste [2002], for example). When there is significant distrust of politicians and parties, a separate constitutional body may have greater popular support (for which reason politicians may not be prepared to go along with such an arrangement).

Box 37. South Africa: One body, two roles

Under the interim constitution, elections were held for a parliament that was also the main constitution-making body. That election was an important milestone; it was the first time most people had voted. It would have been impossible to produce a separate, more legitimate body to deal with the constitution.

This had other advantages. All the newly elected MPs could be involved in constitution- making. Only one secretariat was involved, and only one building needed. It has been observed that although the African National Congress was the dominant party, “the executive did not control the process, as it might have in an ordinary parliament.”

What are the arguments against separating the legislature from the constitutional assembly? Time and expense are clearly relevant; almost certainly a special assembly will take longer than some sort of expert commission process, and probably even than a regular legislative review (although these can also be slow). It will almost certainly be more expensive than if it is parliament that has the role.

From the perspective of interested groups, one disadvantage of a special assembly is likely to be its unpredictability. People who commit themselves to a constitution-making process would like to be able to predict the outcome, in the sense of the constitution they will produce. The more “popular” the composition of an assembly, the less predictable it is likely to be. Designers of the assembly may try to make it more predictable—by bringing in members through political parties, by limiting the voting possibilities of members, and through the internal structures and rules of the constituent assembly. Results of such efforts cannot be guaranteed. Even party discipline may collapse; in the Kenyan National Constitutional Conference it proved impossible for parties to control their members, as ethnic considerations dominated.

Various constituent assemblies have exceeded their mandates. Jon Elster has written:

[T]he [U.S.] framers ignored the instructions from the Continental Congress on three crucial points when they decided to write an entirely new constitution, to seek ratification by state conventions rather than state legislatures, and to require ratification by nine states rather than by unanimity. In France, the constituent assembly decided to ignore the instructions of their constituencies with regard to both the voting procedures and the King’s veto. In Germany, finally, the constituent assembly successfully insisted on ratification by the state legislatures rather than by popular referendum. The German framers also managed to resist some, although not all, of the decentralizing instructions that the Allies had given them (Elster 1995: 375).

Bodies in some processes have been hybrids: legislators are members of the constitutional assembly, but there are many other members. This may seem to be a reasonable compromise, but has its own problems. If the parliamentarians are many, it may be impossible for both bodies to sit at the same time, which may seriously hold up the constitution-making process. They may also have influence greater than their numbers would suggest, raising risks of use of such influence for their own benefit.

Legal status of assemblies

In revolutionary or transitional times, the “legality” of assemblies convened according to the rules set forth in a previous constitution may be doubtful. American commentators have argued

about whether the actions of the constitutional convention were “illegal” and therefore reflected the founding of a new legal order. The French constituent assembly came into existence to raise revenue for the French king. But as the existing system fell apart, it transformed itself into a constitution-making body. Even in this, revolutionary, case, the body itself was convened by the existing authority. In some situations of constitutional crisis, or even vacuum, constitutional assemblies have been brought into being by actions of outside forces, whether in the form of occupying powers, such as the victorious countries occupying (West) Germany after World War II, or the United Nations in Cambodia [1993], or the United States in Iraq [2005]. In Vanuatu [1980], members of the various local parties formed their own constituent assembly, without any formal legal framework, and proceeded to make a constitution. It was accepted by the colonial powers (France and Britain) and became law.

The South Africans, as mentioned earlier, wanted to preserve legal continuity. So the 1993 interim constitution and the final constitution were made by and according to the procedures of the parliaments of the time.

Other assemblies have come into existence in more “normal” times, and have been formed by existing authorities for the specific purpose of making a constitution. Those authorities might be the ordinary legislative process (as in Kenya in the later 1990s) or military authorities (as in Nigeria in 1977 and 1988). In those circumstances it is perhaps easier to design a process and a structure for the assembly. But it is probably rare that a constituent assembly is formed without great pressures from contending forces, even if those forces are not at war. In fact, in less- fraught constitution-making circumstances, perhaps a special constitutional assembly is a less common way to make a constitution. Among the constitutions made or amended in recent times in noncrisis circumstances are those of Finland, Sweden, and Switzerland, where changes were made by the ordinary parliament. This may also reflect national traditions—though Switzerland’s individual cantons do sometimes use constituent assemblies to make their cantonal constitutions.

For the purposes of this handbook, we assume that the important consideration for constitutional process designers will be whether the constitution is acceptable to the nation, something that is likely to depend more on its content and the process of making it than on legalistic arguments about its foundations.


This book is written for those who do have a chance to plan, and there are some lessons that can be learned from the past. At the least, it may be useful to have a checklist of aspects of assembly design. Some main objectives of rational constitutional assembly design are:

  • legitimacy in the eyes of the public;
  • a body that acts on the basis of its perceptions of public interest and not of self-interest (of its individual members, of itself as an institution, of the parties or groups from which the members come or that they represent); and
  • a body that has the necessary competence (including having access to necessary technical assistance).

A purist view would be that a true “constituent assembly,” as the embodiment of the sovereignty of the people, ought to have the entire constitution-making power, subject possibly to the requirement of a referendum. In reality many assemblies have done far less. Past assemblies, on a continuum from widest to narrowest range of functions, include those in table 8.

There is no “right” set of tasks for an assembly. But there are a few considerations that may be helpful in planning those tasks.

On “civic education” (see part 2.2.2), the constitutional assembly may itself need to be educated about its task, and may be poorly equipped to educate the public. And the sort of human, logistical, and other support that the constitutional assembly will have available may not be adequate for the task. However, in South Africa the assembly did take on this role effectively by having its administration hire and train a team of civic educators.

Whether a separate commission or other body should prepare a draft is much debated. One can argue that the people’s sovereignty would be better expressed if the constitutional assembly were to carry out both these operations, because the document prepared in the first stage will almost inevitably impose some restraints on the contents of the final document. If a first draft, prepared by another body, takes account of public opinion, and the document is then revised by a constitutional assembly, it will be, in theory, the outcome of two quite different processes of ascertaining public sentiments (and, if it then goes to a referendum, of yet a third). If the first draft is prepared before the constitutional assembly sits, then one would have to ask when the “political bargaining” would take place: with the drafting commission, or left until the constitutional assembly, in which case is it public or behind the scenes?

Table 8: Functions of constitutional assemblies

Countries Functions
Afghanistan [2004] Approving with little discussion a draft that then requires further act of adoption
Cambodia [1993]; Timor-Leste [2002] Carrying out or supervising the whole process designed by others from draft through promulgation
Nigeria [1979, 1999]; Kenya [2005] Thorough debate on the existing draft, with freedom to change it; further act of adoption by another body
Uganda [1995] Thorough debate of existing draft, but needing supermajority to amend draft; final act of adoption by another body
United States [1787]; India [1950]; Bolivia [2009] Preparation of draft; then full debate; further act of adoption or promulgation required
Vanuatu [1980]; Namibia [1990]; Nepal [ongoing process] Designing the whole process (other than formation of the constituent assembly, and the adoption process) and carrying it out from public consultation (if any) to draft to promulgation

If there is no other body, the assembly may have to be charged with responsibility for public consultation and provided with the necessary staff members. However, this may be a task that an assembly is not well equipped to carry out—the task requires special expertise, including perhaps computer programmers and data entry personnel. (See part 2.3.4 on the personnel needs of a process.) And, especially if the assembly is also the parliament, or parliamentarians are also members, it may be undesirable for them to conduct public consultation meetings: some of the issues the public wants to raise may be concerned with the competence and commitment of legislators generally, or even the particular members at the meetings, and the public would naturally feel hesitant to speak freely.


Constitutional assemblies have a tendency to be large, often too large, though they have varied widely. (See table 9.) If dissatisfaction with parliament is one reason for having a constituent assembly, it will often be felt that representing the whole people requires a larger body. If, as in Kenya, parliamentarians are members of the assembly, the latter must be larger, and there it was decided that others should outnumber members of parliament. Realistically, if a section of society has one member it may feel some sense of satisfaction, but is unlikely to be able to influence decision-making at least by its voting force (except perhaps if a two-thirds vote is on a knife-edge, or if unanimity is required at some point). But there is another role than that of voting: articulating the issues of the various sectors of society.

Large bodies are hard to control. Decision-making is difficult, even in committees because they will also be big, unless they are numerous (in which case providing secretarial and other assistance and keeping track of what the committees are doing will be hard). With large bodies, there is more risk that members will make rhetorical speeches rather than contributions of substance.

Size will be less of a problem if the function of the constituent assembly is mainly to endorse a document prepared elsewhere (such as Afghanistan’s Constitutional Loya Jirga); the body will not really have to understand the document in great detail. Most constitutional assemblies are expected to discuss the issues in detail, and even those that are not expected to may insist on doing so.


A dilemma in constructing any constitution-making body is how to balance the wishes of the people with those of the political classes, especially of political parties. After or during periods of conflict parties may not really represent the people, and the role and control of parties may be an issue in constitution-making. Political parties have interests as parties, and politicians as politicians.

On the other hand, there is a risk that a constitution that is adopted without being accepted by politicians, and even by the organized parties, will not work. A good working relationship between the politicians and the people is essential, and the final document will probably be a product of compromises between these two groups, as well as among other groups in society.

If the legislature is the constitution-making body, it will be composed according to the usual procedures (though if it is elected with a view to constitution-making, the election results may be slightly different than those in regular elections). If the constitution-making assembly is quite separate from parliament, it may seem pointless to use the normal election method, and permit the parties to campaign in the usual way—the composition of the assembly will probably largely mirror that of parliament. However, different people might be interested in standing for elections.

The composition of various constitutional assemblies is shown in table 9.

This shows that some assemblies have been designed to represent the people, sometimes to the exclusion of traditional politicians (who may be otherwise occupied with the regular legislative business, in a separate body). The categories of “people” include, and indeed often emphasize, those marginalized in the past; their demands may be the very basis of the assembly process, as in Bolivia and Nepal. One problem with such a basis is that new groups tend to emerge, demanding their place in the assembly. This was true in Nepal, where the interim constitution had to be amended to make the membership more comprehensive, and the selection process correspondingly became more complex.

Election and selection of members

Assembly members (other than regular legislators) may be chosen by:

  • direct election by the people;
  • indirect election by existing political bodies such as local government councils;
  • election or selection by bodies, not necessarily normally politically active, that would not usually choose representatives to a legislature (such as civil society or parties—other than in a party-list system—or even an individual such as a king); or
  • omination by the chair of the assembly, or some other person or body, for reasons connected with the makeup of the body and its expertise or to repair shortcomings in its representativeness.

The arguments for and against direct election include the following:


  • it is democratic;
  • it can lead to fair representation; and
  • it is likely to be acceptable to people.


  • it is expensive and time-consuming;
  • if the electoral system is faulty (this may be one of the issues to be decided when debating the new constitution), the assembly may not be representative and may have an interest in preserving the faulty system;
  • it may be dominated by political parties;
  • people of competence and integrity may be reluctant to stand for election (or parties may not accept them); and
  • elected members may feel compelled to stick to their mandate and be less prepared to compromise.

Many assemblies are made up by a combination of methods.

Table 9: Composition of constitutional assemblies

Country and body Composition Population of country Assembly size Women
Bolivia [2009] Constituent Assembly Directly elected 9 million 255 35%
Bougainville [2004] Constituent Assembly Members of two existing governmental bodies, 36 in the Interim Bougainville Provincial Government and about 100 in the Bougainville People’s Congress 185,000 136 8
Ecuador [1998] Constituent Assembly Directly elected members 12 million 90 28.5%

Ecuador [2008] Constituent Assembly

Mixed system of some members representing provinces (100) and some citizens overseas (6) and also party lists (24)

13.7 million

130 34%

Eritrea [1997] Constituent Assembly

105 members of the legislature; the rest elected by regional assemblies or selected from representatives of Eritreans abroad

3.2 million

527 30% quota

France [1789] Constituent Assembly

291 deputies of the clergy, 270 of the nobility, and 584 of the Third Estate 28 million 1,145 0
Germany [1949] Parliamentary Council Indirectly elected (state legislatures elected some of their members) 50 million 65 About 4
India [1950] Constituent Assembly Mainly elected by provincial legislatures 345 million About 300 About 15
Kenya [2005] National Constitutional Conference All members of parliament (222): 3 elected by each district council; (no more than one councillor, at least one woman); 126 chosen by civil society; 1 member from each registered party; 13 nominated by chair to represent other aspects of society; constitutional commission members (29) 31 million 629 About 136
Nepal [ongoing process] Constituent Assembly (Constitutional Assembly in Nepali) 240 members elected by first-past-the-post, single-member geographical constituencies (2 of those elected were nonparty candidates); 335 elected on party-list system; 26 nominated through party-based consensus 26 million 601 197
United States [1787] Constitutional Convention Elected or selected by individual states 4 million 55 0

Box 38. Members’ active participation

The Kenyan constitutional commission looked at the contributions of different types of members to plenary debate. In quantitative terms they found that members of parliament made a somewhat smaller contribution than their numbers would suggest, while members of the commission that prepared the constitution draft that formed the basis of discussion contributed significantly more. Only 61 percent of the members of parliament spoke at all, while 91 percent of NGO representatives did, as did 89 percent of representatives of religious organizations. Only 46 percent of those appointed to represent “special interests” did so. These figures probably reflect the lack of belief on the part of many members of parliament that the process would lead anywhere, and the much greater commitment of civil society and district representatives. It was probably a mistake to include members of the drafting commission in the constituent assembly—they did not have a vote, but some had considerable influence over thematic committees.

It has been argued that, whatever the usual electoral system, broad representation of the people in the constitutional assembly would be best achieved by a list system of proportional representation, with a low threshold, to ensure that even small parties are present. In some countries a national proportional representation list system has been adopted (in South Africa, for example), or a list system but using smaller constituencies (as in Namibia). The system adopted for the Nepal constituent assembly was a mixture of single-member geographical constituencies (240 members) and party lists (335 members). It did not rely on the natural tendency of proportional-representation systems to be more inclusive (because people did not trust the political parties to be inclusive in their selection of candidates) but imposed rigid requirements in terms of gender, caste, and ethnicity for the lists and the seats. In Bolivia a party could not necessarily hold all three seats from a district even if it won sufficient votes to do so: if there was another party with at least 5 percent of the votes in the district, it would take the third seat. This was designed to protect small parties. But proportional representation systems do give a great deal of power to political parties.

However, some assemblies have included members drawn from lists not put together by parties. For example, half the lists for the election to the Geneva constituent assembly established in 2008 were party lists, with the other half being lists of people with particular interests (such as homeowners, women, and retired people), including one list put together by a federation of 480 local organizations of all sorts). It is common for a proportional representation list system to provide that a party will only have any members elected as members of parliament if it achieves a certain threshold percentage of the votes cast—at least 1, 3, or even 7 percent. The aim here is to prevent fringe, even extremist, parties from winning seats. But the threshold might be lower for a constituent assembly precisely in order to encourage inclusion of a full range of national interests.

Entitlement to vote in an assembly may be wider than for parliamentary elections. In Eritrea some members of the assembly were elected by Eritreans abroad (though in some other countries, claims by the diaspora of the right to vote have been rejected). Afghans in refugee camps in other countries were also to vote for the Constitutional Loya Jirga. In the case of a country emerging from serious conflict there may be many exiles, but tensions can exist between exiles and those who remained. In Kenya a court held that prisoners could vote in a referendum; they were debarred from voting for parliament. A similar possibility might exist for voting for a constitutional assembly—depending on the words of the existing constitution; if there were no constitution, an expansive view of the right to vote might be possible.

Some other selection process by civil society has been used in a number of constituent assemblies, including Afghanistan [2004] and Kenya [2005]. There are risks in such a system, including those of undue influence or even bribery in “small-circle elections” resulting in new sorts of distorted representation. But if the alternative is that elections are exclusively dominated by unreformed parties, this may be a risk worth running, and in a system where civil society is genuinely responsive to society those risks may not be realized.

Nomination of a small number of members could be designed to bring in representatives of groups and sections of society that are underrepresented or individuals whose experience will add weight to the deliberations or legitimacy to the body. Such a nomination could be made by the chair of the body (as in Kenya, where the chair used the power to appoint some persons from the business sector, including media). But in some instances the chair is not identified in advance. It could be by agreement among parties, as in Nepal, but there the parties to some extent undermined the understood purpose (to bring in expertise without regard to party affiliation) by sharing out the slots among the major parties and not necessarily using them to bring in experts.

Relatively few countries make voting a legal duty. It has been suggested that, even in a country with a practice of compulsion, this should not apply to elections for a constituent assembly. But in New Jersey in 1966 only about 3 percent of the voters voted for the constituent assembly members, which rather makes nonsense of the idea of people’s participation.


If public funding is usually available for election campaigns, arguably it should also be available for campaigns for election to the constituent assembly. But there may be many nonparty candidates in such an election. It has been suggested that there should be a publicly financed information sheet to educate the public about all candidates. A few countries have banned separate party rallies and campaigning, providing only for joint, publicly financed election meetings (e.g., Tanzania in the one-party state period). That plan might be worth considering for elections to constitutional assemblies.

Federations or federating countries

Indirect elections (usually by a lower-level legislature) are common in federating countries, as in the United States at the time of the constitutional convention, and to some extent in India in 1946. The shortcomings of this approach are that:

  • it may make the assembly a process of bargaining among units and not a process of the people; and
  • it may mean that the assembly mirrors distortions in the makeup of the federal state—the very distortions that may be involved in the disputes underlying the demand for a new constitution.

Should representation of units be based on equality or population? The former favors smaller units. Issues of representation should be looked at in the light of voting arrangements, among other things: what number of votes is needed to carry a motion and what number is needed to block one?

Qualifications of members

One might argue that to make the assembly as representative as possible there should be no requirement other than being a citizen, and perhaps a registered voter. Some countries might consider allowing members of the national diaspora to be members of the assembly, even if they are not citizens. They may have left the country against their own wishes, under oppression or conditions of insecurity, and may not have given up aspirations to return to it.

Qualifications imposed in actual cases have included Nigeria’s requirement in 1977 that members be “indigenes” of the state they represented (meaning that their tribe is one that “belongs” in the state). Nigeria excluded traditional paramount rulers in 1988. Both Nigeria and Nepal (2007) excluded current public servants, and Nepal precluded payment to anyone from government funds. Nigeria in 1977 and 1988 required that a person have paid taxes for the previous three years (which might discriminate against women); while Nepal excluded anyone owing the government money. Nepal’s concern to exclude the corrupt led it to bar anyone convicted of crime involving “moral turpitude” or corruption, or of any electoral offense in the previous two years. It went to the extent of barring those blacklisted (under law) as defaulters to banks. And Nepal also barred those under twenty-five years old—thus excluding the voice of youth.

Having a large number of candidates can be confusing for voters. Requirements to show a certain level of support may reduce those numbers. In Nigeria in 1977 the number required was

ten. In Nepal in 2007 only a proposer and seconder were required. Another device for discouraging frivolous candidature is to require a financial deposit. In Nepal an individual standing for a geographical constituency had to put up a deposit of 3,000 rupees (about $45 in United States dollars) and a party putting forward a list for the proportional representation election 20,000 rupees (about $300).

Experts as members?

Elster has suggested that experts should not be members because the constitution should be the outcome of bargaining between the representatives of the people, rather than an expert affair: “Lawyers will tend to resist the technically flawed and deliberately ambiguous formulations that may be necessary to achieve consensus” (Elster 1995: 395). But experts have the same right as any other citizens to stand for election, and this exclusion may produce a distortion of expertise. Some countries have provided for specific seats for persons with certain expertise, while in many others some members will have expertise, with lawyers often being prominent. In India political parties sought to ensure that some leading experts were included in the constituent assembly.

Nonmembers with special relationships

In Kenya [2005], certain observers were permitted to attend, and to interact with members informally during breaks. These were drawn from civil society, and were selected by a committee (from thousands who applied). One principle was to allow groups that had important potential input, but were not formally included in the assembly, to be present. The observers could attend plenary sessions (as could other members of the public) and were assigned to relevant committees (unlike members of the public), but could not speak or vote in either. Twenty-eight organizations were also accredited, which meant they could lobby members, distribute literature, and hold meetings on the grounds of the constituent assembly.

In other contexts, government observers have also been included—perhaps unwisely. In Afghanistan [2004], constitutional commission members who prepared the draft were included as observers along with the transitional cabinet and the heads of the judicial and human rights commissions. They did not have the right to vote or express an opinion unless they were asked a question by one of the members of the Constitutional Loya Jirga.

Payment of members

If constitutional assembly members already hold paid public positions, perhaps as legislators, there is little reason to pay them more for their constituent assembly roles. But in some countries legislators are adept at claiming allowances for everything they do. People who otherwise would not be paid, or not paid much, would have a legitimate claim to be paid for this public service. Ideally, salaries should not be fixed by the members of the constituent assembly themselves. This could lead to abuse, discrediting the assembly.

Structure of the assembly

It seems unnecessary to have a second chamber for a purely constitution-making assembly. It is meant to represent the entire people, and it does so in one chamber. Conflicts with any other body should be avoided. The one purpose that a second chamber could serve would be to prevent a body that is otherwise not subject to any realistic constraints from straying beyond its mandate. To be able to do this, the second chamber would have to have great public legitimacy—which might falter under the strain of disagreement.


It is common for much of the work to be carried out in committees. In addition to committees to discuss substantive issues, there will probably be some addressing administrative matters. Some may resemble the committees often found in legislatures (on the discipline of members, for example). Others may be needed because the assembly is a new body without an existing structure. For example, there may be a committee concerned with the welfare of members, their housing, transportation, and so on. And the special nature of the work of the assembly may necessitate special committees such as those on civic education, the collection of views, and the education of members. Committees have included:

• a capacity-building and resources-management committee and a committee for public opinion collection and coordination in the Nepal process; and

• a staff and finance committee, a credentials (members’) committee, a house committee (concerned with housing, library, and other facilities), a press gallery committee, a steering committee, and an order of business committee (to plan the future course of the constituent assembly) in the Indian assembly.

How should topics be assigned to committees? Sometimes the structure of the existing constitution may be used as the model. This has the disadvantage that the structure of that constitution may be dictated by that of the old, because of the committee structure. Sometimes the committee structure closely reflects national concerns, as it did in the Bolivian constituent assembly, which included committees on “education and interculturality,” “hydrocarbons, minerals, and metallurgy,” “water resources and energy,” “integrated Amazon development,” and “coca.”

The South African assembly was divided into the following committees:

  • Committee I—Democratic state (preamble, citizenship, equality, supremacy of constitution, elections, freedom of information, accountability, separation of powers);
  • Committee II—Separation of powers, legislative procedures, constitutional amendment, structure of government at different levels, legislatures, electoral system, traditional leaders, executive;
  • Committee III—Relationships among levels of government;
  • Committee IV—Fundamental rights;
  • Committee V—Judicial and legal systems; and
  • Committee VI—Public administration, financial institutions, transformation, security services.

The following criteria might usefully be borne in mind:

  • Committees should be no larger than about thirty members if the benefits of having a smaller body are to be realized.
  • Though it is useful to have expertise in committees, it is not practical to give members a free hand to choose their own committees—committees will become uneven in size. Also, it is often important to have nonexperts on committees as well as experts; the whole idea of a constitutional assembly is to have the people’s voice, rather than only experts, involved).
  • The membership of committees should be balanced in gender, ethnic, and other terms, if possible.

If parliament is in charge of the process, the committee structure may be different. Parliament has many other functions to perform, and other committees will exist. Parliament may therefore decide not to assign all its members to the constitution-making process, but to set up one committee with that responsibility. That committee may operate essentially like a constitutional commission.

In Zimbabwe [ongoing process], a parliamentary select committee had an oversight role in the constitution-making process. It appointed seventeen “thematic committees” with 30 percent legislators and the remaining members drawn from “stakeholders” outside parliament.


It may seem natural for the assembly to have its main seat in the national capital for symbolic reasons, and because that is likely to be the most accessible town. There may be good reasons for having it sit elsewhere:

  • to be away from the seat of the national legislature;
  • to symbolize the break with the past;
  • to avoid a place still affected by turbulence; or
  • to avoid the constituent assembly being subjected to undue pressure from the people and organizations of the already powerful capital city.

Similar factors may suggest not using other cities. In Bolivia the constituent assembly sat in Sucre, an old colonial capital, but not the current national capital where the parliament sits. But to some extent the constituent assembly became hostage to the demands of the Sucre people about the city’s constitutional status.

It may also be necessary to strike a balance between protecting the security of the members and ensuring that the security forces are not in a position to overawe them.

Even if the assembly sits in the same town as the legislature, it would be best for it not to sit in the same venue. But it may not be easy to find a place large enough for the assembly, which must have not only a hall for plenary sessions, but many rooms for committees, for support staff and facilities, and (if public access is to be permitted) for public facilities. The Kenyan body sat in a large hall designed for dance performances before tourists (chosen in preference to a city-center conference hall to minimize the risk of delegates slipping away to their offices or to shop). It was not easy to see or hear delegates in the spectator seats; a shortage of rooms was to some extent solved by the use of tents, but this meant that secrecy of deliberations was impossible. In Nepal the use of a conference center meant that the chair was on a stage, and the stage was the only part of the hall visible from the public gallery. In Afghanistan, a large tent was erected.

Support and facilities

The constituent assembly will need:

  • an administrative management body—which will probably require staff members with a wider range of skills than a normal legislative secretariat (see part 3.3 on establishing this body);
  • logistical arrangements for members (possibly including housing);
  • printing facilities of its own, or access to good facilities elsewhere;
  • communications (radio, computers, TV studio, and the like);
  • vehicles or access to vehicles if it is to travel; and
  • security arrangements. (For a more in-depth consideration of what is needed, see part 2.3.)


Predicting how long it will take a constituent assembly to complete its work may sometimes be hard—especially if the assembly has the entire constitution-making role. If it is set up to debate a full existing draft, a clear timetable may be possible—especially by limiting the life of the constituent assembly in order to limit its role.

Time limits prevent groups with more patience (but not necessarily more concern for the national interest, whose interest may be in financial gain) from obtaining advantage. On the other hand, an assembly composed partly of people unused to deliberative work of this type may take some time to understand and feel comfortable with its role. If the process is too rushed, more experienced members (probably seasoned politicians) may have the upper hand.

What happens to an assembly once it has finished its task?

It is important that the law (if any) should be clear about what happens when the assembly finishes its task. The interim constitution of Nepal is a little unclear: “The term of the constituent assembly is to be two years from the first sitting” but “on the day of the commencement of the Constitution promulgated by the constituent assembly” its task is to come to an end.

Turning the constitutional assembly into the first legislature under the new constitution may not be desirable. The assembly may be quite different in its makeup from a regular legislature,

as we have seen. But it is not easy to prevent this by existing law if the assembly is perceived as having full constituent power—in other words, it creates the new legality, which overrides existing law.


The context, as well as national tradition, may often define the choices listed above. Designers of the process need to follow models of other countries with care, and to learn from their own country’s past efforts and experiences.


3.1.3 National conferences

National conferences (sometimes called national conventions) are usually large, unelected bodies composed of representatives nominated by a wide range of interests established to discuss constitutional and other options for the future in situations of intense national crisis in the period of transition from authoritarian (single-party or military dictatorship) regimes to more democratic regimes. They were established mainly (but not only) in French-speaking countries in Africa from the late 1980s to the early 1990s. They were usually established rapidly, in response to a crisis, and often provided the first opportunity in two or three decades for wide- ranging public discussion of the issues facing the country. Their proceedings were generally broadcast live on television and radio.

In addition to government representatives, most national conferences consisted of representatives selected by interest groups such as opposition political parties and civil associations. They tended to be much larger than constituent assemblies; the national conference in the Democratic Republic of the Congo, 1991–1992, had more than three thousand participants. In terms of the constitution-making tasks that they performed, a minority of national conferences developed and adopted a constitutional text, in much the same way that some constituent assemblies do. More commonly, however, the main constitution-making roles involved development of principles and proposals that were shaped into a constitutional text by a transitional legislature established by the national conference. And in part because they often declared themselves sovereign and replaced the existing national executive and legislature, many national conferences had extensive roles beyond constitution-making.

Organizing and managing a conference could be difficult due to the combined effects of such factors as the speed with which conferences were established; the emerging opposition to the existing regime being weak and disorganized in many cases; their large size; and the many groups represented in them.

Origins of national conferences

The origins of most national conferences can be found in intense fiscal and political crises, some involving the collapse of banks; the inability of the state to pay the salaries of public servants; national strikes; and violent clashes of unions and opposition groups with military forces. Such crises have usually been exacerbated by international community pressure—from the mid-1980s, for structural reform of economies, and from 1988 to 1989, for political reform. The extent of the crises meant that the government had little or no legitimacy and so was unable to advance reform through existing institutions. Further, in most cases, political parties had long been banned or were extremely weak, resulting in little support for establishing a new national legislature or an elected constituent assembly as the way forward. There needed to be some other institution that enabled a national government with little legitimacy to engage with a wider set of interests that had not coalesced into a legitimate opposition party or group.

The best-known national conferences are those that were held in French-speaking Africa, the first being in Benin in February 1990. Subsequently they were held in Gabon, Republic of the Congo, Mali, Niger, Togo, Democratic Republic of the Congo, Madagascar, and Chad in the period from 1990 to 1993.

French and other influences

The choice of the national conference as a constitution-making institution in so many countries in crisis in French-speaking Africa was partially influenced by the ongoing connections of those countries with France, which had been the colonial authority in all cases, and continued to maintain close links in most. The significance of an institution that was to be seen as a model for the national conference—the États Généraux, which had met in Paris in 1789 on the eve of the French Revolution—was highlighted in the 1989 bicentennial celebrations of the French Revolution held in France and widely reported in French-speaking Africa. The education of the elites that organized the national conferences in France or in French colonial schools had emphasized the importance of the revolution and the États Généraux.

On the other hand, there were other influences than French links. In particular, large gatherings with some similarities to national conferences were also part of political reform processes developing in other parts of Africa from the late 1980s in countries without a French background, including Ethiopia, Namibia, Somalia, and South Africa. Further, not all French- speaking African countries established national conferences, though there were strong calls for them in the early 1990s from opposition groups in some such countries, for example, Burkina Faso, Cameroon, the Central African Republic, Côte d'Ivoire, and Mauritania.

It is also true that the national conference has not been solely a phenomenon of French-speaking Africa. The first one in Africa was in formerly Portuguese São Tomé and Principe, where a December 1989 conference of about six hundred mainly ruling party members had a limited constitution-making role. (It made recommendations for political reform to the ruling political party.) In part because the possibility of using an institution such as the national conference was opened up as a result of the experience of French-speaking Africa in the early 1990s, countries from other historical and cultural backgrounds have also established bodies with some characteristics of national conferences. Examples include Sierra Leone [1991], Russia [1993],

Indonesia (1999–2002 process leading to amendments), and Rwanda [2003] (though most of these might better be regarded as “hybrid” bodies, as discussed below).

Some features and problems of national conferences

Some closely related features commonly found in national conferences, and some associated issues and problems arising in their operation, require brief comment. They concern:

  • processes for making initial decisions about establishing such a body;
  • the variety of roles and tasks that they carry out, with particular reference to their constitution- making tasks;
  • the large size, the composition by representation of interests, and the short duration of the conferences, and some difficulties associated with those characteristics, in particular

    • difficulties in their acting as effective deliberative bodies; and
    • management difficulties, including translation; and
  • issues about chairing national conferences and the importance of a forward-looking perspective.

All of these features and issues can readily be considered in light of the experience of the nine national conferences held in French-speaking African countries between 1990 and 1993. Some basic information about those national conferences is summarized in table 10.

In addition, brief comments are required on the constitutional outcomes of those same national conferences.

Processes for establishing national conferences

The haste with which crises forced (usually) reluctant authoritarian governments to establish national conferences meant that there was seldom any detailed agreement negotiated between the government and the loose opposition groups about the goals, composition, and operation of the conference. The lack of such agreements made the work of some of the conferences particularly difficult, with government and opposition groups taking antagonistic stands and failing to cooperate. In Togo, soldiers supporting the president surrounded the conference venue on various occasions, enforced a presidential order to suspend proceedings, and later held the transitional legislature hostage.

The main preparatory work was usually limited to establishing preparatory commissions that decided which groups would be represented. In general, they played no role in selecting representatives of such groups, it being left to the groups to decide their own selection methods. Benin was an exceptional case, where a roundtable process was established in advance of the conference in which the government and the main opposition leaders negotiated basic agreements on the process and some initial principles that the constitution to be developed by the conference would follow. In the process, a degree of understanding was established among key actors in the conference, something that probably contributed to the success of that conference compared to most others. (A case study of the constitution-making process in Benin can be found in appendix A.2. Roundtable processes are discussed in part 3.2.2.)

Table 10: National conferences in French-speaking African countries, 1990–1993

Country Dates/duration Participant numbers (approx.) Role: Preparing constitutional principles or proposals, with transitional legislature established by national conference to develop and adopt new constitution Role: Preparing and adopting draft constitution, usually requiring approval by legislature, head of state, referendum, or a combination of these
Benin 1990, 10 days 488 Yes No
Gabon 1990, 3 weeks 2,000 Yes No
Republic of the Congo 1991, 5 months 1,202 Yes No
Mali 1991, 15 days 1,800 No Yes
Niger 1991, 4 months 1,200 Yes No
Togo 1991, 52 days 962 Yes No
Democratic Republic of the Congo 1991-1992, 17 months (intermittent) 3,000+ ? ?
Madagascar 1992, 10 days 1,400 No Yes
Chad 1993, 3 months 830 Yes No

Roles and tasks

National conferences were multipurpose bodies. Concerning their constitution-making tasks, there were two main variations. Most of the conferences developed some proposals about guiding principles for, or recommendations about, the text of a new constitution. Largely because of time limitations, they then usually established a transitional legislature and executive (at the same time ousting the existing executive and legislature) and gave the transitional legislature the task of developing the guiding principles or recommendations into a new constitution. (In practice, the transitional legislatures normally delegated that work to a committee.) In most instances the conference also stipulated that the new constitution be approved by a national referendum. In some instances, as in Chad in 1992, a preparatory commission readied the initial draft for the national conference to consider.

The other main variation involved the national conference developing its own draft constitution (usually together with other basic laws, such as a new electoral law). The short duration of the national conferences meant that in the few cases in which this occurred, most of the work of developing the draft had to be done in committees and working groups. While few developed a draft constitution, there were more that initially decided to take to themselves authority to do so, but when time pressures meant they could not meet for long enough to complete the work, they ended up being among the majority of conferences that delegated their authority to some form of transitional council. Chad was one such case. The national conference had authority to debate, amend, and adopt the draft received from the preparatory commission. But when the process took much longer than expected, the conference delegated the completion and adoption of the draft to a transitional legislature, which then took two years to complete the task.

When a conference did delegate authority over drafting the constitutional text to a transitional legislature, it could nevertheless influence the development of the new constitution in two main ways. The first was in establishing principles and proposals that the transitional legislature was required to take into account when determining the content of the constitution. The second was by determining the membership of the transitional legislature.

In addition to their constitution-making tasks, some national conferences addressed past crimes and human rights abuses. When they assumed sovereignty (as many did), they might also take on responsibility for establishing new transitional governmental institutions and planning elections. Without their functions being clearly defined, it could be difficult for participants to give constitution-making tasks their full attention.

Size, composition, and duration

The size, composition, and duration of conferences varied considerably. Table 10 indicates the wide range of numbers of participants in the nine conferences in French-speaking Africa, from a low of 488 in Benin to a high of more than 3,000 in the Democratic Republic of the Congo. Six of the nine cases had more than 1,000 participants. The numbers are approximate because official membership lists often changed, and because there were often many observers without voting status.

Participants were not selected through elections (as is usually the case with members of a constituent assembly). Rather, they were generally nominated to represent an institution, a political party, or an association of some kind. A huge variety of groups was represented, including labor unions, students’ and teachers’ organizations, human rights groups, professional associations, traditional leaders, religious communities, women’s and farmers’ groups, and educational institutions. The number of parties, groups, and associations represented varied from as few as about fifty (Benin) to about five hundred (the Democratic Republic of the Congo).

Generally speaking, a broadly representative organizing body (usually called a “preparatory commission”) would consult before determining the parties, sectors, and groups to be represented, and the number of seats to be allocated to each. It would then be a matter for each party, association, or group to decide on the method of appointment of the delegates to fill the number of positions allotted to it. Some political parties and associations were newly established in the weeks or months before the conference was set up, sometimes primarily for the purpose of getting conference representation. Jennifer Widner reports that it was easier to select representatives of associations and interest groups in countries where there was a tradition of national level “peak” associations, as in Europe, than it was in countries with a decentralized interest group model of the kind more common in British Commonwealth countries. The former were seen as legitimately selecting representatives, while in the latter cases every small association wanted its own representatives. In general, despite claims of national conferences being representative (in Benin the president claimed that the conference would represent “all the living forces of the nation”), the membership selection process resulted in overrepresentation of the political class and the educated elite in countries where the vast majority of the population was neither (Widner 2008).

The duration of the public sittings of conferences included in table 10 varied considerably, but most sat for much shorter periods than is usually the case with constituent assemblies. Six of the nine sat for periods of three months or less, and two for just ten days. These are remarkably short periods for a body intended to negotiate new constitutional arrangements intended to respond to deep crisis and conflict. Among the reasons for this phenomenon is the large size of the bodies, which can make them difficult and expensive to administer.

Issues arising from large size, complex composition, and short duration

The large size and short duration of national conferences are characteristics that could place obstacles in the way of their being effective bodies to deliberate on major constitutional questions. Such problems could be exacerbated where conferences have not been preceded by roundtable processes encouraging initial understandings between opposing groups. Further, short durations sometimes contributed to a tendency to rush the process, making it difficult for opposing interests to enter real negotiations at the conference. There was pressure on allied groups to make agreements and prepare draft texts in advance, or for the conference to delegate most responsibilities for developing the constitution to another body.

Large size and short duration also tended to reduce the opportunity for careful consideration of and negotiation about difficult and divisive issues. Indeed, it was often hard to organize the time needed for all delegates to speak, although there was usually pressure for that to occur. The more time that needed to be allocated to speeches by individual delegates, the less was available for serious debate about reform proposals. Further, there are particular difficulties in

managing deliberation and negotiation in large bodies, especially where there is a high degree of public scrutiny (as was the case with most national conferences). In such circumstances, with little time to speak, participants are often under pressure to take polarizing positions. The pressure can be worse if elections are likely to follow, as participants may then be under pressure to stake out clear positions that appeal to potential constituents, something that in itself can encourage the adoption of polarizing positions.

The size, composition, and short duration of these institutions can also give rise to difficulties with their management. Locating the funds needed to run large conferences was difficult, especially in situations of financial crisis (so this was a factor in the short duration of the majority of the conferences). Their large size and composition through representatives of groups contributed to their unwieldy nature. Tasks such as the registering of participants were sometimes extremely difficult. It was common for unaccredited participants to slip into parts of the proceedings. Short durations made it difficult to arrange basic services such as translation of documents and speeches, something often critically important in a multiethnic situation. In Chad, a divide that developed in the national conference between French-speaking and Arab communities was made worse by delays in translating proceedings and documents into languages other than French.

Chairing, and looking forward

The chairing of national conferences was critically important. Where chairs were positive, inclusive, conciliatory, and forward-looking, they made valuable contributions. But in circumstances in which the chair placed heavy emphasis on the conference’s role of addressing past crimes and human rights abuses, there was a tendency to look backward, with extensive airing of grievances and demands for revenge, rather than looking forward to what a new constitutional dispensation could have to offer. This was a particular problem in the Republic of the Congo and Togo, and one that was avoided (in large part by effective chairing) in Benin.

Constitutional outcomes of national conferences

While most national conferences did play important parts in processes that resulted in the adoption of new and more democratic constitutions, research by Jennifer Widner shows that the new constitutions offered lower levels of rights protections than countries that used constituent assemblies (Widner 2008). Further, they tended to fail at a higher rate (in the sense that there tended to be a return to higher levels of violence or suspension of the new constitution). Hence there is little evidence that national conferences have had a particularly positive record in terms of outcomes in constitution-making. On the other hand, it is not easy to determine the extent of the contribution to such outcomes of the constitution-making process as opposed to the kinds of economic and political circumstances existing in the countries in question at the time of these constitution-making processes.

Practical tips

This survey of the experience of national conferences gives rise to some practical suggestions for consideration by anyone considering developing an institution such as a national conference:

  • Careful consideration would be required about whether the circumstances are such that a national conference with features similar to those discussed here would be the most appropriate institution for undertaking significant constitution-making tasks. The institution has attractions mainly in a deeply divided crisis situation, where pressures for resolution of the situation are urgent and where there are limited aggregations of interest in established political parties. There are other institutions or procedures that can be used in some such circumstances. One is the roundtable, though it tends to be used where opposition groups are more clearly defined than has usually been the case preceding national conferences.
  • Given some of the difficulties experienced in operating many national conferences, in advance of establishing such an institution, a process similar to a roundtable (see part 3.2.2) should be used in an effort to establish some initial agreement on process and on the principles to be followed by the new constitution, and some political understandings among the parties involved.
  • The size of the membership of the conference should be kept as low as is practicable in the circumstances, as this can reduce management problems and improve the prospects of the conference being effective as a negotiating and deliberative body. Because processes of negotiation and reaching compromise can be undermined by rules and other arrangements under which all members are given time to make opening statements (often creating pressures on speakers to take polarizing positions), it may be preferable to give members opportunities to speak in committees and working groups. Further, if the conference is large, it is necessary to ensure that as much as possible of the negotiation and decision-making about constitutional issues is conducted in broadly representative (and competent) committees.

If associations and political parties are to be represented in a national conference, then to avoid problems with such bodies being established purely for the purpose of representation in the national conference, every effort should be made to agree on basic preconditions to accreditation for representation, including such things as the period for which such a body should be required to exist and the minimum number of members it must possess before it can be accredited.

Every effort should be made to ensure that the chair of the conference is a highly respected figure who is able to keep the work of the conference focused mainly on what is required for development of future arrangements that will reduce conflict and resolve problems, rather than focusing on how to address past abuses by government (there being other processes available to handle past wrongs).

3.2 Institutions that develop proposals about which final decisions are made elsewhere

In this section, the main distinction is between roundtables, which are generally relatively informal bodies, often formed in situation of crisis to put together constitutional proposals, and commissions and committees that are already in existence and are given that same task, or are specially created (usually by law) for that purpose. What they have in common is that they are not like legislatures (though they may be committees of legislatures) because they do not have any power to enact a new constitution.

We have also included here parties to peace processes; we are not concerned with their peacemaking roles, but with those aspects of their roles that relate to the content of a new constitution. It is an important assumption of this handbook that in constitution-making after (or even during) conflict, making a constitution may be an important part of the peace process. Where parties to a conflict are divided over an essentially constitutional issue they are unlikely to give a free hand in constitution-making to a commission or a freely elected constitutional assembly.

As is often the case, the boundaries may not be clearly defined. For example, a roundtable may be a peace process. Because a peace process will usually begin before more formal institutions can be set up, we begin with this topic.

3.2.1 Parties to peace processes

Peace processes intended to end violent conflict often have close links to constitution-making processes, especially when the cause of conflict has been access to state power. Such links can make the parties to a peace process significant, even dominant, actors in the constitution-making process. These parties can have interests and modes of operation different from those of parties involved in the more specialized bodies and processes that usually have the main roles in constitution-making processes.

There can be risks and opportunities in linking conflict resolution and constitution-making. Among the major risks are dangers that, in a situation of violence and insecurity, communities may draw inward, resulting in increased sectarianism, pursuing short-term solutions that protect sectarian interests or that constitutionalize and perpetuate divisions (as, for example, in Burundi). Further, there can be a strong tendency to try to exclude noncombatants. (Some specific dangers involved in that situation are discussed below.)

On the other hand, getting parties to a violent conflict to consider constitutional issues may present opportunities. Where conflict has resulted from perceptions of exclusion, putting the possibility of constitutional change on the table may be one of the few ways to get serious consideration for the idea of ending violence. It does so by creating space for political discussion that may open up possibilities for better understanding of opposing positions and of the possibilities for compromise. Constitutional debate can help opposing parties redefine their concerns. For example, the 2005 Comprehensive Peace Agreement in Sudan redefined the issues from a violently contested demand for Southern independence to how best to meet opposing concerns through constitutional arrangements for autonomy and a deferred referendum on independence. If it is possible to involve leaders of opposing armed groups in discussions with other groups and interests as part of constitution writing (though, as discussed below, this does not always occur), this can expose those leaders to a broader range of needs and concerns, and help moderate polarized positions.

Formal statements of the links between a peace process and a constitution-making process are usually found in the peace agreements made as part of conflict-resolution efforts. The extent to which peace processes have links with constitution-making varies greatly. A peace agreement may set out a road map for the peace process, which often includes the road map for making a new constitution and may not go any further (e.g., toward determining the contents of the constitution). Some peace agreements do deal extensively with constitutional content, in a variety of ways. Some set out guiding or immutable principles upon which the proposed new constitution should be based. Others state quite detailed proposals either for changes to the existing constitution or for a new constitution, usually leaving it to some other body or process to enact the proposals. In a few cases the peace agreement is in fact a constitution, as was the case with the interim constitution of South Africa of 1993. Alternatively, a completely new draft national constitution may be attached to the peace agreement, as with the constitution for Bosnia-Herzegovina attached to the Dayton Accords of December 1995.

Peace agreements are usually intended to be binding on the parties, and so often contain sequencing and other arrangements intended to encourage implementation by the parties (including implementation of provisions on constitutional process or content). In addition, international community actors are often involved in the development and signing of peace agreements (as mediators, facilitators, parties, or witnesses). Such roles may subsequently involve them in encouraging or even actively supporting implementation of the constitutional aspects of the agreements.

The links among conflict-resolution processes and constitution-making in any particular case vary greatly, depending on the nature of the conflict and the goals of the main parties. Such variations can have significant effects on demands about the constitution-making process and the roles parties seek to play in it. Examples of peace processes where key parties may seek to define changes to the constitution but not seek major roles in important aspects of constitution- making include efforts to end secessionist conflicts in which neither the national government nor the secessionist rebels have won a clear victory. In such cases the secessionists may be interested mainly in seeking agreement on the details of constitutional change needed to provide autonomy for the secessionist region, and in a few cases also a deferred right to a referendum on independence, as with New Caledonia in relation to France in 1998, Bougainville in relation to Papua New Guinea in 2001, and South Sudan in relation to Sudan in 2005. In those cases, the peace agreement defined agreed-upon constitutional changes, and left the process for making those changes to the existing national legislature.

The situation can be quite different in cases of conflict involving attempts by a party expressing grievances about previous marginalization and so seeking to capture control of the state. The rebels may be less interested in providing for the content of the final constitution in the peace agreement, and more concerned with obtaining both a commitment to complete replacement of the constitution and a guarantee of a significant, even dominant, role for the rebels in the process of making the final constitution. Where an interim constitution is involved, the rebels can also be expected to have a strong interest in determining its contents. Thus in Nepal (in the case of the Maoists) and South Africa (in the case of the African National Congress), the peace agreements provided for interim constitutions (in South Africa providing guiding principles for the final constitution), which guaranteed the Maoists and the African National Congress significant control of or influence over the interim governments and the processes for making final constitutions.

The different interests and modes of operation that parties to a peace process may have in relation to constitution-making when compared to groups operating within more specialized constitution- making bodies can have significant effects on a constitution-making process arising from the peace process. Several interrelated potential issues (some of which have already been touched upon) require brief discussion, including possible responses to problems that may arise.

First, issues of sectarianism, short-term thinking, and resistance to compromise in situations of violent conflict and insecurity have been mentioned already. The closely related difficulty of limiting participation in the constitution-making process to combatant groups can be a particularly strong tendency when negotiations occur in a situation of ongoing conflict. In such circumstances the parties to the peace process may concur that a lasting agreement will be more likely if confidentiality and secrecy are maintained and the process is controlled by and restricted to the leaders of the warring groups. Further, there could sometimes be risks that opening a negotiating process at an early stage could empower groups and interests without real power or status, and could contribute to substantial elaboration of the agendas of issues. While these may be real dangers that do contribute to the difficulty of the decision-making process, there are usually other important issues involved. In particular, injustices can be done to significant groups that are excluded, something that may exacerbate social divisions. (An example of this is Sri Lanka, where the exclusion of Muslim and Indian Tamil communities from successive peace processes where constitutional issues were a significant part of the process has had ongoing divisive effects.) Limiting those involved in constitution-making decisions to combatants has other dangers. They may lack legitimacy in the eyes of the wider community. The success of any new constitutional arrangements may rely too much on the commitment of the narrow group of negotiating parties, and may fail to respond to broader social needs. The new constitution may then lack the social foundations needed to gain the widespread support it requires to be sustainable.

A second, related set of problems arises because in a constitution-making process linked to conflict resolution, the issue of who has seats at the table will be seen as determining access to power in the long term, through decisions on the constitution. So access to seats in the process becomes critical. The result can be splits in fighting groups, or even the emergence of new groups. Alternatively, because so much is at stake, leaving out any faction can result in the emergence of spoilers.

Problems of both these kinds are sometimes handled in a two-stage process, the first involving mainly the warring parties, which is intended to build confidence and establish order, and the second, more participatory stage, in which the “final” constitution is made. The access of parties to the first stage (usually involving negotiations of peace process matters and short-term constitutional arrangements) can be flexible. Thus in South Africa, the interim constitution (inclusive of guiding principles) was negotiated mainly among the key parties, with some parties joining quite late in the process. The constituent assembly process that developed the final constitution was more participatory. Another approach could involve provision for a mandatory review, within a specified period, of a constitution negotiated between limited parties.

Third, problems can arise from the privileged roles of combatants as parties to the peace process contributing to demands for constitutional provisions that maintain their privileged status under the new constitution. For example, the army or security forces of the state may seek provisions that give them special constitutional roles or protections. Rebel groups may seek incorporation into the state’s armed forces. Another danger concerns a narrowing of the range of constitutional issues that may be considered as part of a constitution-making exercise based on a peace process. There may be a tendency to limit them to issues of major concern to the parties to the peace process, contributing to a lack of balance in the constitution and undermining its wider legitimacy.

Fourth, parties to a peace process can usually be expected to understand well the political context they face, it being that which usually drives them to demand particular constitutional concessions. There is often a tendency to focus mainly on issues about access to political power under a new constitution, and a lack of proper attention to other vital issues. Further, parties may have limited interest in, or even understanding of, the longer-term legal implications of their constitutional demands. Combatant parties to peace processes (especially, though not only, rebel groups) sometimes have limited constitution-making expertise. They may have limited ability to express their grievances and concerns in terms of constitutional issues. Further, they may see constitution-making as just one of many possible strategies for achieving their goals, and keep open the possibility of a return to violence.

As a result, there can be special needs for helping parties to peace processes better understand what is involved in constitution-making. Legal advisers to parties in peace processes may need to be involved early in order to help the parties frame their concerns as constitutional issues, something that can often help redefine issues and transform conflict. (This has occurred in many peace processes, including Bougainville, New Caledonia, South Africa, and South Sudan.) Facilitators of peace processes should give special attention to the need to encourage continued commitment to the constitution-making aspects of the peace process.

A fifth set of issues arises because increasingly, even in peace processes responding to conflicts internal to a particular country, third-party mediators and international actors play significant roles. Often it is their timetable concerns and other agendas that tend to dominate the process. Such actors may be concentrated on resolving the violent conflict, and have a limited understanding of constitution-making processes or issues. These difficulties suggest a need for improved understanding of and training about constitution-making on the part of international actors involved in the many peace processes that are associated with constitution-making.

3.2.2 The roundtable

Roundtables are informal consultative processes sometimes used to negotiate initial steps in a constitution-making process during a period of transition from an authoritarian to a more democratic regime. They usually occur in situations of national crisis, where the existing national constitution does not provide a legitimate basis or adequate guidance for a workable constitutional reform process. Pressure to escape the crisis results in members of the national government consulting the political opposition (and sometimes other interests) about the steps needed to initiate and advance a solution to the crisis, including agreement on constitutional reform that is then usually undertaken in accordance with the requirements of the existing constitution.

The roundtable process therefore usually enables the maintenance of legal continuity, which can be important in situations where those controlling the existing regime remain powerful and would consider a break in legal continuity illegitimate. In this respect, a roundtable is different from many national conferences (institutions also often used in situations of national crisis) where loss of the legitimacy of the existing regime and the extent of the national crisis are often so great that the national conference seeks a break in legal continuity by declaring itself sovereign and establishing transitional constitutional arrangements while a new constitution is being developed. (See part 2.1.9.)

The use of the term “roundtable” in relation to constitution-making processes is a relatively new development. It is most commonly used with reference to the processes in Hungary and Poland in the late 1980s, both part of transitions from authoritarian socialist regimes. (The case study of the constitution-making process in Poland—see appendix A.10—provides an overview of how a prominent example of a roundtable process operated in practice, and also outlines subsequent steps in the constitution-making process in Poland that began with the roundtable.)

The term has also been applied to other processes in Eastern Europe from the late 1980s— particularly those in Bulgaria, Czechoslovakia, and the German Democratic Republic (Eastern Germany)—as well as some in Latin America (notably Chile in 1989 and Colombia in 1990) and to processes in Spain (in 1976) and South Africa (in the early 1990s). Similar arrangements have been used in many other processes without being described as roundtables. Examples include the processes used to establish some of the national conferences (see part 3.1.3) held in French-speaking African countries from 1990 to 1993, the best known example being Benin. (See the case study of the constitution-making process there, in appendix A.2.) The process in pre-independence India from the 1930s for consultation among local actors about decisions on constitutional progress toward independence might also be classified as involving a roundtable.

Roundtables themselves usually involve at least two major steps. First, government and opposition groups engage to decide on structures for negotiations (numbers of representatives, chairing arrangements, working groups, and so forth). Such negotiations can take time—six months in the case of Poland. Second, meetings of the agreed-upon roundtable structures are held, and can often (though not always) be completed in quite a short time—just a few in the cases of both Hungary and Poland. The structures used vary greatly from case to case, there generally (though not always) being a high degree of flexibility in the arrangements, for example concerning criteria for public participation, determining the agenda, and setting decision-making rules. Some roundtables are less flexible. For example, in Hungary in 1989 a three-tiered structure was established. There was also agreement among the nine opposition groups that participated in the process that all their decisions would be made by consensus. One result was that each opposition group had a veto on decisions on joint opposition positions, and hence considerable influence on the roundtable process. Much of the work is usually done in committees and working groups and informal consultations. The work of the roundtable often occurs in secret, or at least without any media or involvement of members of the public.

Because a roundtable has no basis in constitutional or other legal rules, there is usually no hierarchy among the participants, no formal rules for its operation, and no preassigned status even to its most fundamental decisions. When the process starts, the participants will often be quite unclear about the direction in which things will go. The key point is that public participation in the process represents a commitment to negotiating a solution peacefully rather than using the alternative of resort to open conflict and violence.

The main reason why the period within which the roundtable itself meets is quite brief is that the roundtable process is used only to negotiate limited initial steps in a reform process; the legal steps will be taken elsewhere, in institutions with a legal basis. There are cases, however, where roundtable processes take far longer, usually because either the changes being negotiated in the particular case are far-reaching or because the government tends to see itself as still negotiating from a position of strength.

South Africa is an example of far-reaching change being negotiated through such a process. There the process was used to negotiate the concept and content of the interim constitution. The negotiations to establish it, along with the negotiations held to determine transition arrangements, constituted the roundtable phase. The negotiations to establish those processes began in the 1980s, while the negotiations on the arrangements eventually incorporated in the interim constitution were undertaken through the Convention for a Democratic South Africa (CODESA 1 and CODESA 2) and the multiparty negotiating process that ran from 1991 to 1993. In Chile, the government installed by the 1973 military coup continued to see itself as in a fairly strong political position, despite the defeat of the dictator, Pinochet, in a presidential plebiscite in 1988 held under a constitution imposed by the military in 1980. But that unexpected defeat created pressure for political and constitutional change that resulted in almost ten months of intense negotiations among the military government, the main opposition political parties, and political parties that supported the military. Although ultimately successful, the negotiations came close to collapse on several occasions.

After the roundtable has completed its work, the constitutional change agreed to through the process happens elsewhere. In a majority of cases the agreement covers the details of constitutional amendments and also the idea that the amendments will be made through an existing constitutional reform process. That was the case in Chile, Hungary, and Poland, where liberalizing amendments to be made by existing but undemocratic parliaments were agreed upon. Similarly, in South Africa, the contents of the interim constitution were agreed upon, as was the use of the existing parliament to adopt it (thereby providing the legal continuity of such importance to the ruling party). In a few other cases, and especially in preparations for a national conference, a roundtable is less focused on details of constitutional amendments than on the next steps in a reform process. The roundtable process in Benin, for example, did agree on basic principles that should be met by any new constitution, but was much more concerned with agreement on the need for and structures and other arrangements for the national conference.

The high degree of informality and flexibility involved in a roundtable makes it quite different from most peace processes and from legally defined constitution-making institutions such as constitutional commissions and constituent assemblies. There can be advantages to informality (which is often accompanied by secrecy) in that the participants can avoid loss of legitimacy or stature if the process does not achieve particular outcomes. They can then be freer to make some other move to initiate change than would otherwise be the case. There may also be advantages in the flexibility regarding possible outcomes that is permitted by a roundtable process, as this can include arrangements for incremental progress in reforms that can be revisited as the fruits of initial progress are tasted.

On the other hand, where the roundtable marks the beginning of a process that continues to have limited popular involvement (as was the case with the process in Hungary for many years after 1989), secrecy and the lack of public involvement in the roundtable process can contribute to a lack of legitimacy of the constitution resulting from the process.

There can be immense pressures on those involved in attempting to establish or operate within roundtable processes in situations of deep crisis, where there is a grave risk of violence in the event of failure, as was the case in Eastern Europe and Latin America in the late 1980s. Final responsibility rested with those actors. This was a different situation from that of pre-independence India in the 1930s and early 1940s, where most Indian actors felt that the British colonial government had final responsibility, and they could afford to remain relatively disengaged.

Roundtable processes can sometimes offer other advantages in situations involving extremely undemocratic and repressive regimes where inflexible constitutional arrangements have made liberalizing reform difficult to achieve. First, the process can provide legitimacy for a reform process in a situation where the existing regime has little legitimacy, by enabling an inclusive process in a situation where democracy was absent. It allows inclusion of a wide range of actors in a process intended to reach common agreement on the way forward. Second, in situations such as Hungary (and many others) where there is no coherent opposition, a roundtable can provide a framework within which opposition groups can emerge and learn to cooperate, negotiate with the government, and participate in constitutional decision-making, in preparation for participation in government. Third, it can provide a form of limited power-sharing arrangements (between the government and the emerging opposition) where none was possible under the existing constitution. Fourth, it can permit concessions to be made by groups in opposition to the old regime (as was the case in Chile, Hungary, Poland, and Spain) without those concessions necessarily being incorporated into final constitutional arrangements. Instead, the opposition groups remain free to engage in developing more permanent constitutional arrangements when the agreed-upon reforms have worked to change the balance of political forces, thereby opening possibilities for more significant change to occur. Fifth, it can provide for learning (by both the old regime and emerging opposition groups) about constitutions and constitutional limits. Sixth, a roundtable can help develop cooperation among opposing factions (government and opposition) and their leaders that can contribute in many ways to progress in later stages of reform processes.

Practical tips

The following practical tips are offered for anyone considering making use of roundtable arrangements as part of a constitution-making process:

  • A roundtable is mainly restricted to use in situations where there is both an authoritarian regime that retains a reasonable degree of authority but is becoming open to the possibility of reform, and also some coherence and leadership in the opposition groups that enable them to cooperate effectively.
  • It should normally be as inclusive as possible, in terms of elements in the existing regime and the various opposition groups. Only by being inclusive will it give sufficient support to the reforms agreed on through the roundtable process.
  • The practical arrangements for the roundtable process should normally be flexible, and permit the parties to work together in whatever way enables the process to make progress.
  • In addition to working toward agreement on constitutional change and next steps in the constitution-making process, a roundtable should aim to contribute to the development of understanding and working relationships between government and opposition leaders.
  • There are often advantages for opposition groups if a roundtable process agrees to a limited degree of reform that may open the way to more extensive reforms later. There can be a risk for opposition groups that agreement upon too much detail at the roundtable stage may lock in concessions that offer little advantage to them at a later stage of the reform process.