The vast literature on constitutions and constitutional change makes recurrent use of the familiar distinction between flexible and rigid constitutions―while a flexible constitution may be amended in the manner in which ordinary laws are passed, a special procedure or organ is needed for the amendment of a rigid constitution (e.g. Derbyshire & Derbyshire, 1999: pp. 15-16 ; Hague & Harrop, 2004: p. 211 ; Lijphart, 1984: pp. 213-214 ; Lijphart, 1999: pp. 218-222 ; Strong, 1958: pp. 65-66 ). However, while much used and referred to, the distinction between what is flexible and what is rigid is still not very helpful or exhaustive. This is because the great majority of the countries of the world subscribe to a rigid rather than flexible formula: while rigidity is the rule, flexibility is the exception. In consequence, much research efforts have over the years been devoted to the task of operationalizing rigidity and rigidity thresholds and relating rigidity differences to factors that may explain why certain countries have more rigid constitutions than others or may explain how rigidity differences influence other features of political and constitutional life (e.g. Lutz, 1994 ; Lorenz, 2005 ; Melton, 2012 ; Benz, 2013 ; Ginsburg & Melton, 2014 ; Anckar & Karvonen, 2015 ).
Such studies usually engage with decision rules that apply to the final stage of amendment: amendments must be decided by a two-thirds parliamentary majority, by a three-fourths majority, by referendum, by a combination of several such prescriptions, and so forth. The possibilities are multifarious―indeed, Arend Lijphart has made the remark that the countries of the world make use of “a bewildering array of devices to give their constitutions different degrees of rigidity” (1999: p. 218). Much can, however, be added to this bewilderment by expanding research to cover also other stages of decision-making and include, then, other actor and arena constellations like, for instance, what Astrid Lorenz has called “pre-final votes”, i.e. parliamentary and other votes that do not relate to a final document but rather to a declaration of amendment need, a general draft, and the like ( Lorenz, 2005: p. 347 ). The research that is reported here is explicitly in this genre as it introduces an additional dimension to the existing rigidity classifications. Specifically, taking on board the recommendation by James Melton that the proposal process should be taken more seriously by rigidity measures ( Melton, 2012: p. 39 ), the contribution of this study is about rigidity formulas that are applied in the initiating phase of constitutional reform and convey the right of constitutional amendment initiative on specified actors and institutions. The concept that will be used here is “initiative rigidity” (IR for short). About this form of rigidity two research questions are posed here: 1) to what extent do the countries of the world make use of IR; what empirical patterns emerge?, and 2) to what extent does this use of a new indicator moderate and refine rigidity classifications that rely on more conventional methods; again, what empirical patterns emerge? In other words, the research touches in separate as well as combined approaches upon the first as well as later steps in the chain of decisions that bring forth constitutional amendment.
Rigidity data are derived here from close readings of the written constitutions of the countries of the world. The leading source to these documents has been the regularly updated collection Constitutions of the Countries of the World, published by Oceana Publications, New York ( Flanz, various years ), which is an excellent guide to the constitutions and constitutional-like texts from all parts of the globe. Not only contain the editions complete constitutional texts; in several cases the editors also provide expert commentaries as well as useful historical notes and reviews and annotated bibliographies. The year that is targeted is 2012, and for each country the particular release that has been used here is the one latest published up to that year. In terms of structure, the study has four sections. Following this introduction, a second section establishes classifications of the IR-instrument as well as applies the classifications to the constitutional texts. The aim of this section is to convey a description of the varying manners in which IR-devices are used. A third section, also empirical in nature, adds classifications and applications in regards to decision rigidities. The aim of this section is to provide a detailed description of constitutional rigidity methods and to find out to what extent, if any, IR-devices are used to supplement or rather replace more conventional amendment methods. A fourth section, finally, adds conclusive comments.
2. Initiatives: A Bird’s Eye View
For the purpose of this study, upon an examination of the materials at hand, IR-arrangements are divided into three groups which denote differing degrees of rigidity. The first group contains cases which do not in fact employ IR-devices, and therefore represents non-rigidity. The bulk of these cases are countries, and Antigua-Barbuda, China, Lesotho and Singapore may be given as random examples, which have not introduced in their constitutions separate IR-prescriptions. Evidently, in such cases, legislation concerning constitutional amendment is initiated in the same manner and by the same methods that are applied to ordinary legislation. Also in this first group, there are a few cases where IR-instructions are given, but it is clear from the context that the instructions do not deviate from the ordinary pattern ? this is the case, for example, in the Marshall Islands, where the constitution states that amendments shall originate in the Nitijela (Parliament), and “shall be considered and disposed of as if they had been proposed by Bill” (art. XII; section 1-3). In all 82 countries out of 193, the percentage mark being 42, are in this first group. Roughly speaking, then, the pattern is that prescriptions on IR are given in the constitutions of well over a majority of the countries of the world. The amount is certainly impressive and serves to indicate that the IR-institution receives much attention world-wide and merits comparative study and examination.
The imminent task is now to bring order and rankings to the initiative policies of the universe of the remaining 111 countries that make up the above majority. This is done by means of a typology which introduces a rigidity distinction between two groups of countries. The typology is given in Table 1, and builds upon three dichotomized dimensions, which are, for reasons given below, regarded relevant to the purpose at hand. The dichotomization is for each dimension about rigidity vs. super-rigidity, which means that the obstacles to the exercise of the right to initiative are few and surmountable as against many and impassable. In acknowledgement of the many difficulties and pitfalls that usually append to attempts in the social sciences at weighting variables and constructs (e.g. Foweraker & Krznaric 2000: p. 766 ), the three dimensions are not weighed,
Table 1. Initiative rigidity in constitutional amendment: eight types.
and they therefore carry the same weight in terms of classification. The overall configuration is therefore quite straight-forward and calls for no more than simple addition―whenever a given country attains a super-rigidity score (SR) on two or three dimensions out of three, this country is taken to represent a super-rigidity class of amendment initiative. Correspondingly, whenever a country attains a rigidity score (R) on two or three dimensions out of three, this country is taken to represent a rigidity class.
The first of the three dimensions that define the typology is about the number of instances and authorities that are given in the constitutions the right of initiative. These are in the following called “actors”, and they may be individual (His Majesty, the President, any member of the National Congress, etc.) or collective (the Government, the Senate, Parliament, citizens totaling at least five percent of the electoral rolls, etc.) in nature. The guiding assumption is that a small number of actors (few) imply a higher extent of rigidity than a larger number (many) which opens up more windows of opportunity. Or, quite simply, if several actors are given the right of amendment initiative, this is a more open, flexible and easy procedure than if the initiative is the privilege of a select few actors only. In the following, the dividing line between few or many actors is set, somewhat arbitrarily, at one or two actors on one hand (few) and more than two actors on the other hand (many). Examples of countries in the first category are Algeria, where revisions shall be undertaken on the initiative of the President of the Republic (art. 174) and Belgium, where the federal legislative power has the right to declare that it is necessary to amend a constitutional provision as it designates (art. 195); one further example, now identifying two separate actors, is from the Constitution of Nicaragua, where it is said that proposals for amendment may be tabled by the President or by one-third of the members of the National Assembly (art. 191; Wiener, 2007: p. 665 ). One example of a less rigid text, inviting several initiative-holders, may be picked from the Constitution of Brazil, stating (art. 60) that amendments may be proposed by at least one third of the members of the Chamber of Deputies or the Federal Senate, by the President of the Republic, or by more than one-half of the Legislative Assemblies of units of the Federation, each manifesting its decision by a simple majority of its members.
The second dimension is about the extent of support among initiative actors for the initiative in question―obviously, this is a requirement that concerns collective actors only. The guiding principle is simple enough, as it states that a demand for an established backing of an initiative renders the amendment procedure difficult and rigid whereas a less demanding approach identifies an easier and less rigid method. The operationalization of this threshold focuses on whether or not majorities are prescribed―if this is not the case, the country in question is regarded as having a less rigid initiative method than countries that have established majority thresholds. For instance, in Cameroon amendments may be proposed either by the President or by Parliament, when signed by at least one- third of the members of either House (art. 63; Taku, 2007: p. 161 ); since initiative powers are in this case assigned to a single person or to a group that does not necessarily constitute a majority, Cameroon is on this dimension classified in a less rigid category. On the other hand, in Bhutan the initiative requirement is for a simple majority of the total number of members of parliament (art. 35), and in Sao Tomé and Príncipe the Constitution may be revised by initiative of three-quarters of the Deputies to the National Assembly (art. 122; Beyer, 2007: p. 790 ). Since these and similar cases lay down a majority threshold, they are classified here as super-rigidity instances; the same goes for a few cases in which the majority requirement applies to one only of two or more collective actors. Also, in a few cases, majority stipulations are for a half rather than half plus one―for instance, in Eritrea proposals for amendment may be tabled by the President or by 50 percent of all the members of the National Assembly (art. 59; Hagos, 2007: p. 294 ). Again, these cases are here taken to satisfy the demand for a majority threshold, and they are classified accordingly.
In like manner, the third dimension is about the easiness or difficulty of access to the initiative institution. The dimension is about the extent to which voters and the citizenry at large have the right of initiative. Some countries have installed mechanisms to that effect―for instance, in Peru, besides the President and Congress, amendment initiatives may be launched by “a group of citizens equivalent to three-tenths of one percent of the voters” (art. 206), in Romania, amendments may be initiated by at least 500.000 citizens who belong to at least half the number of counties in the country (art. 150; Tanase, Pasoi, & Dimitriu, 2007: p. 755 ), and in Palau amendments may be proposed by a Constitutional Convention and the Parliament, but also by petitions signed by not less than twenty-five percent of the registered voters (art. XIV; Ottley, 2007: p. 699 ). Such countries are regarded here as rigidity cases; other countries, who are in lack of such mechanisms and therefore exclude initiatives from the public at large, are classified as adherents rather to the idea of super-rigidity. The introduction of this dichotomy goes back on a framework that identifies and regards as problematic attempts on the side of political elites, the classe politique, to create a coalition and to obstruct attempts on the part of voters and citizens to create a political agenda which is against the will and ambitions of the elites. Adherents to this view often assert that newly entering legislators and parties soon belong to the political establishment and quickly realize that the benefits and costs have become different from the non-politicians; they also emphasize that the coalition of politicians against the voters is not restricted to the politicians in power, but covers the whole classe politique and includes public administrators who serve the politicians ( Frey, 1992: pp. 216-217 ). Anyhow, the rationale for the analysis here is that the opening up of the constitutional initiative to include the general public is a more generous and therefore less rigid method than the reserving of the right of initiative to the classe politique only.
The distribution of countries on categories is presented in Table 2, which is an empirical follow-up exposition of the theoretical framework which was given in Table 1. While certainly comprehensible in terms of logic, some figures in Table 2 still appear artificial as they stem from correlations that are rather technical in nature. For instance, the fact that citizen engagements appear to correlate with the right of many actors rather than few to initiate amendment is not very surprising, as the acceptance of a citizenry engagement almost by necessity works to enlarge the actor sphere. Anyhow, from Table 2 a handful of conclusions follow:
1) The number of countries that represent SR-rigidity clearly exceeds, perhaps somewhat surprisingly, the corresponding number of countries with R-rigidity only. The actual numbers are 72 as against 39. In fact, slightly less than 40 per cent of all countries in the world represent SR. Given the earlier observation that a good 40 per cent of all countries are in a non-rigidity category, the resulting pattern looks two-tailed, as the extremes are well-represented, and the intermediate section is fairly small. In short: four countries out of ten relinquish IR-stipulations; another four out of ten apply a rigorous rigidity code; two out of ten content themselves with fairly mild instructions only. As evident from the figures, the profiles of the 72 countries that have opted for an installation of the super-rigidity device are heavily dominated by two combinations out of four possible. About half of the countries represent an approach where prescriptions for few actors only combine with majority requirements; about one third represent this same combination and add also a denial of the right of citizen involvement.
2) There are fairly obvious differences in terms of frequency and emphases within the three dimensions that make up the general typology, and the differences are in line with the above inclination towards an austere interpretation of rigidity. One conspicuous observation is that the right of initiative rests with few actors in close to two thirds of all cases as against many actors in close to two fifths of all cases. Interestingly, geography appears to explain some of this difference. Reflecting, presumably, a predominance of presidential regimes and hierarchical political structures more generally, three fourths of the countries in
Table 2. Initiative rigidity in constitutional amendment: eight types; number of countries.
Asia and in fact more than four fifths of the African countries and almost all countries in the Middle East have adopted a model of few initiators only. On the other hand, reflecting, presumably, a more pluralistic view of relations between political elite actors, the countries in the Americas and especially in Europe have attained a more balanced appreciation of the models of few as against many initiators―in fact, in Europe 16 cases as against 14 have preferred the multi-actor model.
3) Majorities are requested more frequently among countries that accept few initiative actors only (32 cases out of 69) than among countries that allow many actors (13 cases out of 42). In itself, this pattern is perhaps somewhat surprising, as one would have expected the acceptance of many initiators to be to some extent counter-checked and neutralized by an introduction of majority requirements. This is, however, not the case. The introduction of majority requirements is rather used to accentuate further the rather tight and restricted approach to IR-power that is evident in the allowance of few initiative actors only. Interestingly, testing the geography dimension does not now result in much, as countries tend to behave more or less similarly wherever they are on the globe. The avoidance of majority prescriptions is evident in Africa, where three fifths of the IR-countries waive such prescriptions, and even more so in American countries and in European countries, where close to three fourths and slightly more than three fourths respectively act in like manner.
4) In regards to the citizen initiative device, it appears that the use of this device reflects rather well a more general inclination of countries of the world to make use of or abstain from the use of direct democracy. About one quarter of the countries that have IR-stipulations have extended to the citizenry the right of initiative; this proportion is in approximate correspondence with the observation that few democracies made more than occasional use of the referendum device in the final quarter of the twentieth century ( Hague & Harrop, 2004: p. 162 ). Of European nations, 9 have established the citizen initiative device in the materials at hand whereas 22 have not; this is more or less in agreement with the notion some years ago from a survey of European Union countries which suggested that the level of direct democracy influence was fairly strong or fairly good in 8 countries as against modest or weak in 18 countries ( Kaufmann, 2004: p. 27 ). Finally, of microstates with populations of one million or less, one may find in the materials at hand four states that have an IR-arrangement whereas 12 have not; the proportion does not differ much from corresponding findings from other studies which state that the vast majority of microstates and small island states in the world are in fact non-users of direct democracy devices ( Anckar, 2004: pp. 380-384 ). Indeed, one is left with the impression that the willingness to accept citizen initiatives follows in quite many cases from a cultural openness to public opinion and participation.
3. Correcting for Initiative Rigidity
Since a majority of the countries of the world make use of IR, the thought comes natural that efforts at measuring and evaluating constitutional rigidity should preferably aim at incorporating IR with the methods hitherto used in rigidity research. This section of the paper at hand aims at elucidating and penetrating this thought further; specifically, the section attempts to explore by means of a modest empirical investigation the extent to which the observing of IR really makes a difference. In this investigation, for samples of countries, rigidity rankings that do not involve IR-measures are compared to rankings of the same countries that are based on IR-performance also. The strategy is as follows:
Two sets of countries are identified; within each set the countries are selected to be more or less equal in terms of a constitutional rigidity that does not involve IR. In the first set are twenty cases that are randomly selected from all cases that have a fairly modest score of 2 (two) or 3 (three) on an available rigidity scale that runs from 0 to 9 ( Anckar, 2013 ). In the second set are another twenty cases that are selected from all cases that have a more demanding score of 4 (four) or 5 (five) points on the same rigidity scale. Let it be added that most countries of the world are placed within a scale range that runs from 0 to 5; very few cases, indeed, represent higher scores ( Anckar, 2013: pp. 177-178 ). In the list that is given in Table 3 the respective countries are now assigned also the IR-values that have been established in this investigation and were reported in Table 2, this meaning that the countries are classified as lacking in IR-arrangements (0); as representing a somewhat moderate IR-rigidity (R), in which case they are given one additional rigidity point; or as representing cases of a fairly decided IR-su- per-rigidity (SR), in which case they are given two additional rigidity points. One example may be given from Table 3, which marks original scores of the respective countries by the letter X, and new rigidity scores after adding IR-marks by the letter Y. In this Table Bhutan has an original score of two (2) points, but receives after considering IR two additional points. They are given in due of the fact that motions for constitutional amendments in that country shall be initiated by a simple majority of the total number of members of Parliament (art. 35), an arrangement that merits for inclusion in the super-rigidity category. However, before embarking on a further analysis of scores and patterns, a presentation of the above-mentioned original rigidity scale that forms a sort of backbone for analysis must be inserted. Space does not allow a detailed presentation; a review of main points only must suffice.
Basic rigidity scores are assigned to combinations in the amendment process of legislative and popular sovereignty, the first concept denoting the impact of legislative and executive institutions and the second denoting the extent to which constitutions make use of the referendum device either as an optional alternative or as an absolute requirement. Concerning legislative sovereignty, no rigidity points are scored for arrangements in which parliament may take amendment decisions by simple majority. Arrangements requiring two-thirds majorities (or equivalents) are given 2 (two) points and decisions that require three-fourths majorities (or equivalents) are given 3 (three) points. Countries that make use in constitutional amendments of mandatory or facultative
Table 3. Constitutional rigidity performances of 40 countries: comparing rigidity scores before and after adding IR-scores.
Note: X = rigidity score before adding of IR-score; Y = rigidity score after adding of IR-score.
referendums score an additional two points; if prescriptions for a qualified referendum majority vote apply, the point input is raised to three. Such countries, let it be added, are many in number, as about half of the countries of the world maintain in amendments the one or the other form of the referendum device ( Anckar, 2014: pp. 13-15 ). Refining the analysis of legislative sovereignty, notions of repetitions (same instance, several sittings) and veto-players (several instances), extended sets of scores are manufactured. Joint sittings of two houses or any similar arrangement motivate a modest writing up of the rigidity score (0.5 point); if separate supportive sittings are required, the writing up is the same 0.5 point. If in these contexts qualified majority or supermajority requirements apply, writings up are with the same coefficients that apply for parliamentary decisions. Any introduction of noticeable delay mechanisms between validating readings of amendment bills (e.g. Ghany, 2013: p. 27 ) calls for 1 (one) additional rigidity point. Furthermore, submission procedures, implying that one instance may refer a decision to another instance are awarded 1 (one) rigidity point―one example is Togo, where the amendment requirement is for a four-fifths parliamentary majority but the President may still always refer an amendment proposal to popular vote (art. 144; Adjovi, 2007: p. 922 ).
A specific problem arises from the fact that equally valid prescriptions may point in different directions. They may in fact entail a higher rigidity threshold, as they, for instance, given certain circumstances and conditions, imply the additional use of referendum, or they may, on the contrary, imply a lower threshold, as they, given certain circumstances, create a possibility for avoiding a referendum. Djibouti, France and Ivory Coast may be given as examples. In Djibouti amendment decisions are as a rule taken by a simple parliamentary majority and thereafter submitted to referendum; however, the referendum procedure may be dispensed with a decision of the President in which case the bill for amendment shall be approved by a two-thirds parliamentary majority (art. 91; Rahe, 2007: p. 258 ). In France, again, amendment decisions are by majorities in both Houses and an accompanying referendum; however, the President may avoid a referendum phase by resubmission to Parliament which must now decide by a three fifths majority (art. 89; Lay, 2007: p. 323 ). Finally, in Ivory Coast amendment requires a two-thirds parliamentary majority and an accompanying referendum; however, for most matters the referendum stage may be avoided if the President decides to submit the matter in question to parliamentary consideration only, in which case a four-fifth parliamentary majority is required (art. 125-127; Maledje, 2015: pp. 31-32 ). It is simply the case that amendment procedures may prescribe alternate ways and methods, and this does not necessarily constitute a problem when and if the research task is the classification of methods per se. Usually, however, as in the case at hand, comparative politics research tasks are about classifying countries in terms of methods―then, when countries use alternate methods, problems arise. The solution to this problem that is applied here is to accept the one alternative that stands for the highest level of rigidity. If in any given country two or more alternative amendment methods are given, they are classified separately in terms of rigidity and the most rigorous one is taken to represent the country in question.
Anyhow, Table 3 and Table 4 report the empirical grounds for analysis. Table 3, as noted, lists the 40 countries that are included in the investigation and reports for each country the actual sum of rigidity points, when IR-scores are added to the original rigidity scores. Reworking these data, Table 4 reports in the form of a simple matrix the spreading of countries over a rigidity universe once IR-notions are introduced. The overall impression is that the consideration of the proposal stage alters to a noticeable degree the relative positions of states and that the alteration is on the increase the more the original scores picture rigidities. The average score in the group of countries that originally scored 2 or 3 rigidity points is 2.40 points but increases upon inserting IR-marks to 3.10 points; the corresponding increase in the group of countries that originally scored 4 or 5 rigidity points is from 4.55 to 5.65 points. In short, the comprehensive view is that the rigidity scores of close to half of the countries do not change upon the introduction of the initiative stage, this meaning, of course, that these countries do not as a rule have in their constitutions special directions for amendment proposal. On the other hand, the introduction of the initiative stage implies in one fifth of the cases a modest increase of one rigidity point and in the remaining about one third of the cases a more evident increase of two points. Among countries in the first group are Andorra, Chile, Georgia and Poland; among countries in the second group are Benin, Cape Verde, Chad, Mongolia, Palau and others.
The proportions change to some extent when and if the two groups of low and high rigidity countries are observed separately. Of the twenty countries that originally scored 2 or 3 rigidity points, no less than three fifths remain in the same group and position even after the introduction of IR-marks, while the remaining countries move to higher positions. Of these countries a few advance one step only in terms of total rigidity, while a majority show a more considerable increase. On the whole, then, there is much stability in this group, but there are also evident streaks of mobility and change. As one moves up the rigidity ladder, still more mobility enters the picture. Of the twenty countries with original rigidity scores of 4 or 5 points, about one third keep an unchanged position whereas two thirds now register higher and much higher scores. In fact, no less than one fifth of the countries in this group register two point increases. A random but illustrative example sheds further light upon the traffic from and
Table 4. Constitutional rigidity performances of 40 countries before and after adding IR-scores: mobility matrix.
between groupings that follows in the wake of introducing IR-scores. Observations on traditional rigidity scores only suggest that Armenia, Cape Verde, San Marino and Vietnam display similar and fairly low rigidity marks and therefore represent an internally coherent group, whereas Bangladesh, Bolivia, Kiribati and Ukraine represent another internally coherent group with a clearly higher rigidity profile. The introduction of IR-marks, however, clearly disrupts this pattern. Armenia and Cape Verde both move to the second group and now join Bangladesh and Kiribati in that group, from which Bolivia and Ukraine, in turn, release themselves to join other countries in still higher rigidity categories.
As it involves efforts at evaluating and contrasting various constitutional mechanisms, the art of measuring rigidity is trying and calls for careful consideration―quoting James Melton, identifying an optimal amendment procedure “is extremely difficult” ( Melton, 2012: p. 3 ), and creating a measure that captures the rigidity of constitutional amendment procedures “is difficult, to say the least” ( Melton, 2012: p. 39 ). The attempt that has been made here to shed some light on amendment initiatives has certainly testified to the need for expanding the field of rigidity analysis. In sum, it is evident from the above calculations that the accounting for IR makes in many cases a difference that alters the rigidity profiles that emanate from more traditional approaches to rigidity. The amplitude of change is of course dependent on the nature of the apparatus that is used for measuring initiative rigidity, and one may perhaps suggest that the method that has been used here of assigning one point to prescriptions for a rigid and two points to prescriptions for a super-rigid initiation is insensitive and too mechanical and may exaggerate the actual difference between rigidity systems. A suggestion worth considering implies a division into halves of the proposed notations, meaning that rigid proposal methods would authorize one half and super-rigid proposal methods would authorize one rigidity point. The counterargument is that the proposed grading fits after all rather well in the comparative framework that has come to use in this exercise. In reason, a rigid proposal stage constitutes a more demanding device than, say, a requirement for a joint sitting of bodies (which, as noted, authorizes one half rigidity point), and a super-rigid stage is a more demanding device than, say, an introduction of delay mechanisms of some sort (which, as noted, authorizes one rigidity point). Be this as it may, the overarching finding of this study is that measurements of constitutional rigidity should preferably observe not only constellations and structures of the decisive amendment stage but should also include the methods that are used for proposing amendments. These methods do tighten and supervise the access to the amendment process and by that they certainly affect and add to the degree of amendment difficulty.