The architecture of emergency constitutions

The architecture of emergency constitutions Abstract Nine of ten countries currently have emergency provisions written into their constitutions, here simply referred to as emergency constitutions. The nature of these provisions remains poorly understood. We therefore aim at providing answers to two questions: (i) how much additional discretionary power do emergency constitutions allow and which political actors are given the additional power; and (ii) is there a limited number of “typical” emergency constitutions that combine various aspects in similar or even identical fashion? To answer the first question we construct an Indicator of Emergency Powers (INEP) which takes six central elements of emergency provisions explicitly into account. To answer the second question, we draw on cluster analysis and identify six well-defined clusters. Both the INEP as well as the six clusters allow us to answer important follow-up questions such as what the factors are that determine a country’s choice of emergency constitution but also under what conditions governments are likely to declare a state of emergency given the prevalent emergency constitution. 1. Introduction Today, some 90 percent of all constitutions worldwide contain explicit provisions for how to deal with states of emergency.1 The inclusion of emergency provisions—those legal rules specifying who can declare an emergency, when they can do so, and what actors have what powers once it has been declared—into constitutions has become the norm. They are used quite frequently and their use is often far from innocuous. Between 1985 and 2014, at least 137 countries declared a state of emergency at least once, thereby creating situations with particular political powers.2 Under a state of emergency, some individual rights and liberties are usually suspended and the separation of powers is curtailed in favor of the executive or even a single person like the head of state or government and, by implication, to the detriment of parliament and the courts. Assuming that states of emergency can be crucial moments for the development of entire countries and taking into consideration how frequently they are used, it is amazing how little we know about constitutional emergency provisions. Little is known about (i) the amount of additional powers granted to governments acting under a state of emergency, (ii) the trends in the evolution of emergency provisions over time, (iii) the factors that cause societies to adopt them in the first place, and (iv) their effects, i.e. the effectiveness in reaching the goals stated in the underlying legislation. Here, we focus exclusively on the first two questions, leaving the analysis of the latter questions to follow-up research. Cross-country studies analyzing the additional powers that emergency provisions grant their respective governments as well as the development of emergency provisions over time are extremely scarce. The existing cross-country studies rather focus on their consequences. Davenport,3 e.g., is interested in the determinants of state repression and asks whether the kinds of citizen rights mentioned in the constitution are significant for explaining variation in repressive behavior across governments. In that context, he asks whether explicitly listed limitations on citizen rights play a role and takes into account whether “martial law” and/or “state of emergency” are explicitly mentioned in the constitution. He further asks whether restrictions on either martial law or state of emergency are found. He is thus interested in neither the precise architecture of emergency constitutions nor their possible determinants. Keith and Poe4 ask whether emergency clauses are effective and are, hence, also concerned with their consequences and not their architecture.5 Due to this gap in the literature, we describe the main components of emergency constitutions as well as how they have changed over time. To make the powers that emergency constitutions allocate to governments comparable over time and countries, we develop an additive Index of Emergency Powers (INEP) that consists of six separate subindices. Quite often, legal scholars assert that emergency constitutions do not lend themselves to meaningful comparison. If this were true, questions dealing with the effectiveness of emergency constitutions would be almost impossible to answer. In this article, we ask whether this assumption is true and draw on cluster analysis to see whether we can find a limited number of clusters that share a number of traits. Drawing on 351 constitutions (both current and defunct) for which we have sufficient, available information and relying on thirty-one different variables capturing the most important features of emergency constitutions, we identify six such clusters and, hence, show that emergency constitutions often share many similarities. We find preliminary evidence that the consequences of emergency constitutions differ along the lines of these clusters. This result can now be used to ask a number of follow-up questions such as whether a particular type of emergency constitution is generally more effective than others, or whether we can identify specific situations in which a particular type is doing better than the other types and so on. The rest of the article is structured as follows: Section 2 names the main features of emergency constitutions and identifies time trends in their evolution. Section 3 briefly describes the main features of cluster analysis whereas Section 4 reports the results of the cluster analysis and a set of simple descriptive statistics. Section 5 concludes and spells out a number of follow-up questions that can be tackled with the INEP and the clusters here proposed. An online Appendix contains five tables which offer additional information such as precise definitions of the variables used, their sources and descriptive statistics. 2. Trends and emergency powers We define an emergency constitution as the set of formal legal provisions encoded in the constitution that specify who can declare an emergency, under which conditions an emergency can be declared, who needs to approve the declaration, and which actors have which special powers once it has been declared that the constitution does not assign to them outside emergencies. What we refer to as the “emergency constitution” here is, hence, not a document separate from the ordinary constitution but those formal provisions of it that explicitly deal with emergencies. We therefore do not capture decrees promulgated by government during a state of emergency considered to be part of the emergency constitution, conferral of emergency powers in statute, or any informal constitutional conventions or political tradition. Consequently, we consider the formal provisions literally written in the constitution as the outline of a legal maximum of government actions that are constitutionally allowed for. As such, emergency constitutions are paradoxical documents: Their declared goal is to re-establish constitutional order by temporarily suspending it. They are also paradoxical in the sense that the constitution spells out the conditions under which its regular application may be suspended. As such, emergency constitutions deal with the delicate balance between suspending individual rights by temporarily reducing the separation of powers, while also providing monitoring mechanisms intended to reduce the likelihood that the state of emergency is misused by power-maximizing politicians. Recent examples of countries calling a state of emergency include France, following the various terrorist attacks in Paris in November 2015, the May 2016 emergency declaration in Venezuela, and the October 2016 declaration in Ethiopia. These recent cases can serve to illustrate how emergency constitutions are applied and the large differences in which powers are available to the executive during emergencies. The French emergency constitution gives the authorities the right to derogate a number of basic rights, including the right to censor the press, yet the additional right mostly used has been the right of the police to search private homes without a warrant. Conversely, when Venezuelan president Nicolas Maduro declared a state of emergency in May 2016, he further restricted press freedom. When the government of Ethiopia in October 2016 declared a state of emergency following very large protests against the government, it also severely restricted information freedom. In both cases, the action was in violation of the countries’ own emergency constitutions, and the declaration in Venezuela ought to have been approved by the legislature before taking effect. In this section, we therefore start by giving a general overview of constitutional emergency provisions. We first look at their emergence and diffusion over time. We then deal with six central components of emergency provisions in detail. 2.1. On the diffusion of emergency constitutions Constitutional emergency provisions arguably do not make sense under absolute monarchy or totalitarian regimes. If the head of the executive is unconstrained, why should there be special provisions giving him powers he already enjoys? This is why the history of constitutional emergency provisions is closely linked with the advent of constitutional monarchy and with restricted government more generally. France was the first modern nation state to introduce constitutional emergency provisions in 1795.6 Emergency constitutions have since spread into those countries whose legal development was heavily influenced by France. In 1808, Joseph Bonaparte became king of Spain and the corresponding constitution—named after the French city of Bayonne because it was negotiated there—included explicit emergency provisions as in the French model. The provisions contained in the Bayonne constitution in turn served as a model for many Latin American constitutions. In fact, with the exception of Portugal (1826), all of the subsequent countries to create emergency constitutions in the following decades were Latin American. In chronological order, these were Argentina (1819), Chile (1822), Brazil (1824), United Provinces of the Rio de la Plata (1824), Bolivia (1826), Peru (1826), Ecuador (1830), Uruguay (1830), and Venezuela (1830).7 Two observations are noteworthy: first, in all of the cases mentioned, the inclusion of emergency provisions into the constitution was part of an entirely new constitution and, hence, not part of constitutional amendment.8 Second, all of these countries belong to the French legal family.9 By 1850, twenty countries had an emergency constitution; today, all of them are coded as “French legal origin.”10 But the spread of emergency constitutions did not stop in Latin America or countries of French legal origin. Using the extensive information contained in the Comparative Constitutions Project (CCP),11 we find that by 2013, 171 countries had in fact adopted emergency constitutions.12 This development is clearly illustrated in Figure 1, where we plot the share of independent countries with an emergency constitution, starting in 1900. The figure shows relative stability before 1950—the changes are mainly due to constitutions being abolished such that countries are coded as having no emergency constitution in the interim between abolishing the old constitution and either implementing a new constitution or re-instating the old. After 1950, two developments have clearly affected the share: the independence of former colonies, starting in the late 1950s, and the de facto independence of countries formerly part of or controlled by the Soviet Union. Figure 1. View largeDownload slide Share of constitutions with emergency provisions, 1900–2013. Figure 1. View largeDownload slide Share of constitutions with emergency provisions, 1900–2013. The data show that once a country has included emergency provisions into its constitutions, it is unlikely ever to get rid of them.13 Most countries that did get rid of their emergency constitution—often as a consequence of abolishing the constitution altogether—introduced a new one after a short interval. The only exception is Austria, which abolished its 1934 emergency constitution and has not re-introduced another to date.14 In all other cases, the abolishment was the consequence of a military or communist takeover that eventually resulted in the re-introduction of emergency provisions. As just described, the first constitutional emergency provisions in the nineteenth century were all part of entirely new constitutions. Virtually all of today’s emergency constitutions were already included in the last “new” constitution of a country. The 2002 amendment to the constituton of the Czech Republic is a notable exception in which emergency provisions were included later on.15 Hence, having an emergency constitution is now the rule rather than the exception. Yet, as we outline next, there is considerable variation with regards to specific details. 2.2. Single components of emergency provisions We suggest that every emergency constitution must, at least implicitly, deal with six different questions: a. What are the necessary conditions for a state of emergency? b. Who has the power to declare a state of emergency? c. Who has the power to declare the end of an emergency? d. Who has the power to monitor the legality of the means used during a state of emergency? e. Who exercises emergency powers? f. What (additional) competences does a state of emergency confer to the emergency government? (a) Necessary conditions Concerning the first dimension, the necessary preconditions for a state of emergency to be declared, two trends are noteworthy. First, emergency constitutions have become broader in the enumeration of events that can justify the declaration of an emergency. We rely on the variable in the Comparative Constitutions Project,16 which lists six possible preconditions for declaration: (i) war/aggression, (ii) internal security, (iii) national disaster, (iv) general danger, (v) economic emergency, and (vi) threat to constitutional system.17 Any of the given aspects might be found in a given proportion of constitutions, i.e., the share of constitutions including them can assume a value between 0 percent and 100 percent in any year. Adding up the resulting proportion for each of the six categories can, hence, yield a theoretical maximum of 600 (six categories of 100 percent). Albeit a crude measure, it does give an impression of how one aspect of emergency constitutions evolved over time: For 1950, the actual sum was 97.14 whereas the same exercise for 2011 yields a sum of 167.53. In other words, the possible causes for declaring a state of emergency have been considerably broadened and the typical, modern emergency constitution mentions two. Second, Table 1 shows that some preconditions have become more widely included than others. In 1950, 7.14 percent of all constitutions named any kind of “national disaster” as a potential justification for declaring a state of emergency, this proportion had grown to exactly one-third by 2011. “Economic emergencies,” on the other hand, were explicitly mentioned in 5.71 percent of all constitutions with emergency provisions in 1950. By 2013, this proportion had increased to only 7.73 percent. The sixth precondition in the list given above, “threat to the constitutional system,” did not exist in 1950, but by 2013 it was included in 4.12 percent of all constitutions with emergency provisions. Table 1. Percent of all emergency constitutions that name the respective topic as possible reason for calling a state of emergency. Topic 1950 2011 War/foreign aggresion 31.9 48.6 Internal security 31.9 38.8 National disaster 6.9 26.2 General danger 16.7 25.7 Economic emergency 5.6 7.1 Threat to constitutional system 0.0 2.7 Topic 1950 2011 War/foreign aggresion 31.9 48.6 Internal security 31.9 38.8 National disaster 6.9 26.2 General danger 16.7 25.7 Economic emergency 5.6 7.1 Threat to constitutional system 0.0 2.7 View Large Table 1. Percent of all emergency constitutions that name the respective topic as possible reason for calling a state of emergency. Topic 1950 2011 War/foreign aggresion 31.9 48.6 Internal security 31.9 38.8 National disaster 6.9 26.2 General danger 16.7 25.7 Economic emergency 5.6 7.1 Threat to constitutional system 0.0 2.7 Topic 1950 2011 War/foreign aggresion 31.9 48.6 Internal security 31.9 38.8 National disaster 6.9 26.2 General danger 16.7 25.7 Economic emergency 5.6 7.1 Threat to constitutional system 0.0 2.7 View Large (b) Power to declare The second component of emergency constitutions deals with the question of who has the power to declare an emergency. No matter how precisely the necessary preconditions are defined, some actors need to decide whether they are present or not.18 On the one hand, one can imagine a constitution that allocates the power to declare an emergency to the head of the executive branch without any other organ needing to approve of this decision. This would be equivalent to very few checks on executive power. On the other hand, an alternative scenario would be that more than one constitutional actor must be involved in the declaration and more than one actor must approve it.19 A third scenario, located somewhere between the two more extreme approaches just mentioned, could be that some other branch needs to be consulted; if its advice is not followed, this might increase the political costs of declaring an emergency in terms of reduced state legitimacy. Ideally, the necessary majorities should also be taken into account.20 Emergency constitutions typically reduce the degree of the separation of powers for a limited period of time. Yet, regarding non-emergency periods, there has been a secular trend toward a stricter separation of powers. The variable CHECKS, which proxies the degree of checks and balances and is part of the Database of Political Institutions,21 had a mean score of 1.2 in 1975 which had risen to 2.8 in 2012 (around a stable median of 2). Likewise, the mean score of Henisz’s22 PolConIII measure of political veto player power approximately doubled from 0.16 in 1950 to 0.3 in 2012, the latest year for which data is available. It seems possible then, that even emergency constitutions place more checks on the emergency government than they used to, if they follow the general constitutional trend. In the following we document that, in fact, the opposite has in general happened. Regarding the competence to declare a state of emergency, we observe two somewhat contradictory trends. In 1950, six of ten (22/37 = 59%) countries with an emergency constitution gave the head of state the right to declare a state of emergency. By 2011, this proportion had increased to 80 percent (129/159 = 81%) as most constitutions in the newly independent countries after the de-colonization wave of the 1950s and 1960s gave declaration rights to the head of state. This indicates the overwhelming importance of a single person, namely, the head of state. On the other hand, around 10 percent (16/159 = 10%) of all constitutions currently containing an emergency constitution allocate that competence to the entire cabinet. This is no increase in the separation of powers between legislature and executive, but at least an increase in the number of actors involved in the decision. Keith23 goes one step further and asks whether the responsibility for declaring a state of emergency is given explicitly to the legislature. In 1979, the first year of her dataset, this was the case in 4.2 percent of all countries. By 2010, this proportion had increased to 10 percent. If legislatures declare a state of emergency and executives are in charge during emergencies, then this change denotes an increase in the separation of powers. Another way to implement checks in emergency constitutions is to require the declaration of a state of emergency to be approved by an actor other than the one declaring it. In 1950, 44 percent of all emergency constitutions contained such a provision. By 2011, this proportion had increased to 56 percent, again indicating some increase in the level of checks. Today, consent of the following organs is most frequently required: (i) the first (or only) chamber of the legislature (39 percent); (ii) both chambers of the legislature (19 percent) and (iii) the government/cabinet (14 percent). Approval provisions are important because states of emergency have often been misused by self-serving politicians. In particular, some of them have simply dissolved the legislature to eliminate a watchdog. As of 2009, 15.5 percent of all emergency provisions explicitly excluded the possibility of dissolving the legislature during a state of emergency. In 1979, the respective number was only 3.9 percent.24 According to the CCP, no emergency constitution implemented before 1950 included an explicit ban on disolving the legislature. (c) Power to end The third important component of emergency constitutions is the power to end a state of emergency. One possibility—famously used by the Romans—is to have it expire automatically (after six months in Republican Rome). Here, too, the separation of powers is central. Acknowledging that every state of emergency entails the danger of misuse and the possibility of developing into a permanent autocracy, it seems reason able to allocate the power to end an emergency to an actor other than the one endowed with the exercise of emergency powers. Ackerman25 proposes a “supermajoritarian escalator”: the longer the state of emergency lasts, the more inclusive the parliamentary majority necessary to sustain it. Put differently, over time, ever smaller factions of parliament can end the state of emergency. Keith26 asks whether the emergency is constrained to a set time period and whether an extension is subject to legislative approval. In 1979, a little less than 19 percent of the 153 surveyed constitutions had such provisions, in 2009 the proportion had increased to 35.9 percent. At least de jure, many emergency constitutions thus try to sustain a high degree in their separation of powers by making extensions dependent on the consent of the legislature. (d) Power to monitor legality of actions under emergency Fourth, regarding actors who could monitor the legality of the means used under emergencies, both the legislature and the judiciary seem natural candidates. Ferejohn and Pasquino27 identify a “legislative model” of emergency powers which they distinguish from the (neo)-Roman model. In the legislative model, the legislature “is expected to monitor the use of emergency powers, to investigate abuses, to extend these powers if necessary, and perhaps to suspend them if the emergency ends.” They quote France as an example of a presidential system in which the legislature can impeach the president if it assesses that he has overstepped his powers (art. 16 of the French Constitution). As an example of legislative monitoring under a parliamentary system, they mention independent commissions in the UK installed to monitor the executive in the way it implements anti-terror legislation. Independent courts are a possible ex-post monitoring device. The US Supreme Court decisions Ex parte Milligan28 and Ex parte Korematsu29 are two (in)famous examples. Yet, many scholars traditionally believe that speed is of the essence in emergency situations and that judicial review should be postponed.30 Others, such as Ackerman31 or Dyzenhaus,32 point at the dismal record of the judiciary in constraining government action under emergency. Unfortunately, we are not aware of any dataset that would allow us to identify any time trend regarding this aspect.33 (e) Who exercises power? It seems natural to think of the head of the executive branch as the actor exercising emergency powers. Other provisions are, however, possible. They include the head of the military, but also technocrats. The French version of a state of emergency, the état de siège, implies an expansive delegation of powers to the military.34 Again, we are not aware of any dataset that would allow us to identify any change in this aspect over time. Finally, one must ask what competences are conferred onto the emergency government. First, emergency governments frequently entail the competence to suspend a number of basic rights. The proportion of countries whose constitutions provide for the suspension of rights during a state of emergency has remained virtually unchanged at 70 percent (30/43) in 1950 and 69 percent (118/172) in 2011.35 The dataset compiled by Keith36 analyzes the same issue from the opposite angle: according to her, the proportion of countries that do give a list of non-derogable rights or include a statement that certain rights cannot be infringed has changed from 5.2 percent in 1979 to almost 26 percent in 2010. Beyond the suspension of rights, there is a vast heterogeneity in the competences conferred on the emergency governments. Everything from “all powers necessary” to very detailed enumerations exists. Provisions frequently found include measures (i) to keep all government organs broadly conceived in office (vote of no confidence impossible, all elections suspended, terms of constitutional court judges extended), (ii) to increase the size of the army (including the use of national guards for military purposes), and (iii) to keep the country solvent (introducing new taxes, levying payment of existing taxes in advance) but also more dubious measures such as exempting state servants from all legal liability of state acts committed under martial law or enabling forced labor. Of course, there are also constitutions that expressly limit the competence of government under states of emergency, e.g., spelling out that no constitutional amendment can be passed under a state of emergency or that all decrees issued are only valid until the end of the state of emergency. 2.3. A new power index In order to synthesize these different aspects into a single dimension, we develop a measure that can be thought of as capturing the difficulty—or political cost—of calling and maintaining a state of emergency as well as its potential benefits. The simplest way to create an Index of Emergency Powers (INEP) is to rely on variables proxying for the most important aspects just described and add them up. Using the extensive information available from the Comparative Constitutions Project, this is exactly what we do. In the following, we therefore employ information covering up to 351 current as well as defunct constitutions for which we have been able to obtain sufficient information. Although the CCP naturally covers many more constitutions, central information of specific features in many now defunct constitutions is simply missing, which limits our sample to the present 351 examples.37 The INEP takes into account (i) the degree to which the right to declare a state of emergency is concentrated in a single person—or very few—or limited by multiple veto players; (ii) the need to and the degree to which this right is concentrated; (iii) how many different situations are explicitly mentioned in the constitution and can be used to justify the declaration of a state of emergency; (iv) whether fundamental civil and political rights can be suspended during a state of emergency; (v) whether parliament can be dissolved during a state of emergency; and (vi) whether the government can introduce censorship of the media and expropriate property during an emergency. The first three variables are, hence, concerned with the rules for declaring a state of emergency, whereas the last three are concerned with the powers that government enjoys under a state of emergency. The first three can also be thought of as “the cost element” of declaring a state of emergency whereas the last three cover “the benefits element” of running a state of emergency from the point of view of the incumbent government. In the INEP, higher coding in general implies more power to the executive. This is manifested in a lower degree of separation of powers captured in the first three components (the cost part of the INEP), where a high code implies a relative ease in declaring an emergency. For all elements of the index, no limits on the respective aspect is coded 3, some as 2, uncertainty as 1, and tight limits as 0. As we scale each of the six separate components on a 0–1 scale and subsequently scale the entire INEP on the same 0–1 scale, a coding of 1 would imply that there are no effective limits to the powers of the executive during emergencies and a coding of 0 that limits are maximally tight. The entire INEP is an additive index of the six separate components, as the different parts all represent different mechanisms allowed by the emergency constitution through which governments and leading political actors can directly affect decisions during emergencies. As such, a given level of checks and balances can, for example, be achieved by either limiting declaration power or approval power or perhaps including a sunset clause or a substantial limit on the powers given to the executive during an emergency. In other words, there is no clear progression across the six components, which would have necessitated a non-additive construction of an INEP. As such, due to the lack of a fine-grained theory that would inform us about the relative importance of each of those six components and their interplay, it seems straightforward to simply add the components up.38 In Table 2, we summarize the construction of the six components of the INEP. The components capture the power to declare a state of emergency and how concentrated it is, approval powers, conditions that are progressively more inclusive or vaguely defined, whether or not the legislature can be dissolved during emergencies, whether or not basic rights can be suspended during emergencies, and whether or not the constitution allows for expropriation of property and censorship.39 The INEP thus broadly consists of two main dimensions: (i) a cost dimension consisting of the first three components that outline the political costs and difficulty of declaring an emergency; and (ii) a benefit dimension that outlines the particular, discretionary powers that are allocated to the executive (or other government actors) during an emergency. These include the possibility to derogate basic rights, introduce censorship and expropriation without due compensation, and—for the executive—to dissolve parliament. Table 2. Constructing the INEP. Component Additive coding based on: Declaration power 2    if declaration rights rest with the head of government or the incumbent government; 1 if they are vaguely defined; 0 if they rest with the legislature or other (mainly courts) Approval power 3    if emergencies need no approval; 2 if approval rights rest with the head of government or the incumbent government; 1 if they are vaguely defined; 0 if they rest with the legislature or other (mainly courts) Conditions 3    if conditioned on “internal security” or “general danger”; 2 if they include “economic emergency” or “constitutional threat”; 1 if they include “other” or are vaguely defined; 0 if conditions are only “war” and “natural disasters” Dissolution power 1    if parliament can be dissolve during emergencies; 0 otherwise Rights suspension 3    if all rights can be suspended during emergencies; 2 if some can be suspended; 1 if the provisions are vague; 0 if no rights can be suspended Expropriation and censorship 1    if censorship can either be introduced during emergencies or is constitutionally allowed; 2 if authorities can expropriate without due compensation during emergencies Component Additive coding based on: Declaration power 2    if declaration rights rest with the head of government or the incumbent government; 1 if they are vaguely defined; 0 if they rest with the legislature or other (mainly courts) Approval power 3    if emergencies need no approval; 2 if approval rights rest with the head of government or the incumbent government; 1 if they are vaguely defined; 0 if they rest with the legislature or other (mainly courts) Conditions 3    if conditioned on “internal security” or “general danger”; 2 if they include “economic emergency” or “constitutional threat”; 1 if they include “other” or are vaguely defined; 0 if conditions are only “war” and “natural disasters” Dissolution power 1    if parliament can be dissolve during emergencies; 0 otherwise Rights suspension 3    if all rights can be suspended during emergencies; 2 if some can be suspended; 1 if the provisions are vague; 0 if no rights can be suspended Expropriation and censorship 1    if censorship can either be introduced during emergencies or is constitutionally allowed; 2 if authorities can expropriate without due compensation during emergencies View Large Table 2. Constructing the INEP. Component Additive coding based on: Declaration power 2    if declaration rights rest with the head of government or the incumbent government; 1 if they are vaguely defined; 0 if they rest with the legislature or other (mainly courts) Approval power 3    if emergencies need no approval; 2 if approval rights rest with the head of government or the incumbent government; 1 if they are vaguely defined; 0 if they rest with the legislature or other (mainly courts) Conditions 3    if conditioned on “internal security” or “general danger”; 2 if they include “economic emergency” or “constitutional threat”; 1 if they include “other” or are vaguely defined; 0 if conditions are only “war” and “natural disasters” Dissolution power 1    if parliament can be dissolve during emergencies; 0 otherwise Rights suspension 3    if all rights can be suspended during emergencies; 2 if some can be suspended; 1 if the provisions are vague; 0 if no rights can be suspended Expropriation and censorship 1    if censorship can either be introduced during emergencies or is constitutionally allowed; 2 if authorities can expropriate without due compensation during emergencies Component Additive coding based on: Declaration power 2    if declaration rights rest with the head of government or the incumbent government; 1 if they are vaguely defined; 0 if they rest with the legislature or other (mainly courts) Approval power 3    if emergencies need no approval; 2 if approval rights rest with the head of government or the incumbent government; 1 if they are vaguely defined; 0 if they rest with the legislature or other (mainly courts) Conditions 3    if conditioned on “internal security” or “general danger”; 2 if they include “economic emergency” or “constitutional threat”; 1 if they include “other” or are vaguely defined; 0 if conditions are only “war” and “natural disasters” Dissolution power 1    if parliament can be dissolve during emergencies; 0 otherwise Rights suspension 3    if all rights can be suspended during emergencies; 2 if some can be suspended; 1 if the provisions are vague; 0 if no rights can be suspended Expropriation and censorship 1    if censorship can either be introduced during emergencies or is constitutionally allowed; 2 if authorities can expropriate without due compensation during emergencies View Large Aggregating such issues into one index shows that emergency powers were quite limited in most constitutions for which we have information that were introduced before the Great Depression. A number of constitutions implemented in the 1930s, conversely, gave substantial powers to the executive, for instance, those of Yugoslavia (1931, INEP = 0.50), Poland (1935, INEP = 0.53), and Brazil (1937, INEP = 0.55). During the post-World War II period, as we illustrate in Figure 2, one can also observe clear developments. Several newly independent countries, for example, introduced emergency constitutions with strongly limited discretionary rights but rapidly increased those rights in amendments and new constitutions. Ghana’s 1957 constitution, with an INEP of 0.14, did not leave any declaration rights with the government and clearly delimited both approval rights and the conditions under which emergencies could be declared. Based on its 1969 constitution, Ghana’s INEP score increased to 0.51. The new constitution allocated all declaration powers to the head of government, allowed quite a few conditions under which to declare an emergency, and allowed basic rights to be suspended during emergencies. The evolution of Ghana’s constitution is, overall, representative of the general development of former colonies as they became independent and implemented constitutions of their own. Figure 2. View largeDownload slide Average INEP, 1950–2010. Figure 2. View largeDownload slide Average INEP, 1950–2010. While there was a slight tendency for emergency constitutions to be more delimiting during the late 1980s and 1990s,40 in more recent years a number of new constitutions allowed for substantial unchecked emergency powers to the executive. New constitutions in Kenya (2010, INEP = 0.55), Guinea (2010, INEP = 0.66), and Hungary (2011, INEP = 0.71) all leave significant discretionary power to the executive during broadly defined states of emergency. These developments can be clearly seen as an uptick in Figure 2.41 Although they seem to coincide with the time after the 9/11 attacks, it is unclear whether the attacks caused the uptick in the figure as it is driven by constitutional changes in countries not directly affected by terrorist threats. Separating the six components of the INEP, as we do in Figure 3, reveals that the main development in the early years after World War II is a concentration of declaration rights. Similarly, the figure shows that the new constitutions, implemented in former colonies between 1960 and the late 1970s, also were more likely to allow for the dissolution of the legislature. Conversely, constitutions implemented after the collapse of communism clearly tend to be less likely to allow expropriation and censorship but likely to include more conditions under which a state of emergency can be declared. Figure 3. View largeDownload slide Six separate indices, INEP, 1950–2010 Figure 3. View largeDownload slide Six separate indices, INEP, 1950–2010 Overall, as we show in Table 3, there is no substantial difference between democracies and non-democracies regarding emergency powers—using the DD (democracy and dictatorship) index indicator developed by Cheibub et al.,42 the average INEP value is 0.31 in both groups. The DD data also allow for the separation of three types of democracy (parliamentary, presidential, and mixed) and three types of dictatorship (civil, military, and royal). In the data, only royal dictatorships in our sample have much stronger declaration rights—other autocracies do not—while presidential democracies tend to include more conditions under which a state of emergency can be declared. Historically mixed democracies, i.e., democracies with mostly ceremonial presidents without actual powers, have been less prone to allow suspension of rights, expropriation and censorship, but in the present emergency constitutions, all types of democracies now appear similar on average. Table 3. Characteristics of emergency constitutions, regime types. Democracy Autocracy Parliamentary Mixed Presidential Civil Military Royal Overall INEP 0.32 0.26 0.33 0.29 0.33 0.30 Declaration power 0.66 0.71 0.69 0.71 0.78 0.85 Approval power 0.21 0.18 0.19 0.21 0.23 0.17 Conditions 0.29 0.24 0.44 0.23 0.30 0.21 Dissolution power 0.18 0.13 0.08 0.09 0.09 0.03 Rights suspension 0.19 0.08 0.25 0.15 0.18 0.08 Expropriation and censorship 0.39 0.25 0.32 0.36 0.43 0.46 Democracy Autocracy Parliamentary Mixed Presidential Civil Military Royal Overall INEP 0.32 0.26 0.33 0.29 0.33 0.30 Declaration power 0.66 0.71 0.69 0.71 0.78 0.85 Approval power 0.21 0.18 0.19 0.21 0.23 0.17 Conditions 0.29 0.24 0.44 0.23 0.30 0.21 Dissolution power 0.18 0.13 0.08 0.09 0.09 0.03 Rights suspension 0.19 0.08 0.25 0.15 0.18 0.08 Expropriation and censorship 0.39 0.25 0.32 0.36 0.43 0.46 View Large Table 3. Characteristics of emergency constitutions, regime types. Democracy Autocracy Parliamentary Mixed Presidential Civil Military Royal Overall INEP 0.32 0.26 0.33 0.29 0.33 0.30 Declaration power 0.66 0.71 0.69 0.71 0.78 0.85 Approval power 0.21 0.18 0.19 0.21 0.23 0.17 Conditions 0.29 0.24 0.44 0.23 0.30 0.21 Dissolution power 0.18 0.13 0.08 0.09 0.09 0.03 Rights suspension 0.19 0.08 0.25 0.15 0.18 0.08 Expropriation and censorship 0.39 0.25 0.32 0.36 0.43 0.46 Democracy Autocracy Parliamentary Mixed Presidential Civil Military Royal Overall INEP 0.32 0.26 0.33 0.29 0.33 0.30 Declaration power 0.66 0.71 0.69 0.71 0.78 0.85 Approval power 0.21 0.18 0.19 0.21 0.23 0.17 Conditions 0.29 0.24 0.44 0.23 0.30 0.21 Dissolution power 0.18 0.13 0.08 0.09 0.09 0.03 Rights suspension 0.19 0.08 0.25 0.15 0.18 0.08 Expropriation and censorship 0.39 0.25 0.32 0.36 0.43 0.46 View Large Tracing the development of emergency constitutions over time, we thus find both general patterns and substantial differences across countries. Emergency constitutions have become more prevalent while the discretionary powers that they confer on the executive continue to vary. 3. Cluster analysis In the last section, we mentioned that emergency constitutions first evolved in France and then spread across the countries belonging to the French legal family. Some scholars propose that the type of emergency constitution brought about within the French legal family is different from the one brought about within the common law legal family. The French approach is referred to as the state of siege (état de siège) whereas the British one would originate from martial law.43 Two aspects in particular make them distinct from each other, namely, (ii) the identity of the actor authorized to monitor the state of emergency and (ii) the degree of judicialization. Under a state of siege, monitoring would primarily fall to the legislature, whereas it would fall on the judiciary, and more precisely the regular courts, under martial law. An important consequence of this difference is that monitoring by the legislature can take place during an emergency while monitoring by the courts will only take place after the state of emergency has been ended. Other than for the state of siege, martial law could be characterized by “the absence of statutory foresight for its initiation and use.”44 This last trait makes the identification of martial law via the analysis of formal constitutions somewhat difficult. We now ask whether this dichotomy regarding types of emergency constitutions is reflected in the data, i.e., whether there are clear “families” of emergency constitutions. In order to assess whether one can identify families of emergency constitutions, we employ cluster analysis. To generate clusters, we rely on exactly the same variables used for creating the INEP. Cluster analysis is a set of methods that, at its most basic, aims at minimizing the distance in n-dimensional space between member observations within a cluster while maximizing the distance between clusters. In other words, cluster analysis is a way to identify families of observations that are as similar as possible while being distinctly different form other families. We use the K-means technique with random centroids from which a K-means algorithm optimizes the Euclidean distance within and between clusters.45 As we do not have strong priors about the correct or merely reasonable number of clusters, we perform a set of analyses defining between two and ten clusters. Based on this set of analyses, we select what we believe is the number of clusters that best fits the data. Our criteria are that (i) intra-cluster distances ought to be relatively small—i.e. we require clusters to be fairly coherent; (ii) distances between clusters ought to be comparatively large; and (iii) no single cluster has disproportionately high intra-cluster distances or large outlier members. This last criterion ensures that we do not obtain clusters that are “residual” clusters only containing observations that do not fit into any other cluster. We employ these criteria using graphs similar to standard screen plots in which the gain in identification can be easily eyeballed. In the following, we outline the details of these analyses before describing our preferred solution. 4. Are there typical emergency constitutions? 4.1. The number of constitutional clusters In Table 4, we outline some pertinent features of each of the first nine K-means cluster solutions we obtain. Table 4. Cluster solution specifics. No. of clusters Average intra-cluster distance 2 3 4 5 6 7 8 9 Cluster 1 3.05 2.92 3.31 3.05 2.81 2.22 2.09 2.58 Cluster 2 3.33 3.11 2.88 2.88 2.37 2.53 2.74 1.89 Cluster 3 3.18 2.69 2.24 2.31 3.27 2.29 2.69 Cluster 4 2.92 2.92 2.11 2.39 2.62 2.65 Cluster 5 2.97 3.19 3.44 2.22 2.95 Cluster 6 3.49 2.25 3.46 4.91 Cluster 7 3.09 2.44 2.11 Cluster 8 3.05 2.43 Cluster 9 3.56 “Goodness of fit” measures Av. distance 3.19 3.07 2.96 2.81 2.72 2.74 2.61 2.83 Min. distance 3.05 2.92 2.69 2.24 2.11 2.22 2.09 1.89 Max. distance 3.33 3.18 3.31 3.06 3.49 3.44 3.46 4.91 Largest outlier 8.012 8.34 7.72 7.35 7.13 7.25 7.54 7.36 Montenegro (2007) Germany (1949) Hungary (2011) Germany (1949) Germany (1949) Germany (1949) Hungary (2011) Montenegro (2007) No. of clusters Average intra-cluster distance 2 3 4 5 6 7 8 9 Cluster 1 3.05 2.92 3.31 3.05 2.81 2.22 2.09 2.58 Cluster 2 3.33 3.11 2.88 2.88 2.37 2.53 2.74 1.89 Cluster 3 3.18 2.69 2.24 2.31 3.27 2.29 2.69 Cluster 4 2.92 2.92 2.11 2.39 2.62 2.65 Cluster 5 2.97 3.19 3.44 2.22 2.95 Cluster 6 3.49 2.25 3.46 4.91 Cluster 7 3.09 2.44 2.11 Cluster 8 3.05 2.43 Cluster 9 3.56 “Goodness of fit” measures Av. distance 3.19 3.07 2.96 2.81 2.72 2.74 2.61 2.83 Min. distance 3.05 2.92 2.69 2.24 2.11 2.22 2.09 1.89 Max. distance 3.33 3.18 3.31 3.06 3.49 3.44 3.46 4.91 Largest outlier 8.012 8.34 7.72 7.35 7.13 7.25 7.54 7.36 Montenegro (2007) Germany (1949) Hungary (2011) Germany (1949) Germany (1949) Germany (1949) Hungary (2011) Montenegro (2007) Note: average, minimum and maximum distances reported in the lower panel are inter-cluster distances. The largest outlier is the distance of the given country to the centroid of the cluster in which it is placed. View Large Table 4. Cluster solution specifics. No. of clusters Average intra-cluster distance 2 3 4 5 6 7 8 9 Cluster 1 3.05 2.92 3.31 3.05 2.81 2.22 2.09 2.58 Cluster 2 3.33 3.11 2.88 2.88 2.37 2.53 2.74 1.89 Cluster 3 3.18 2.69 2.24 2.31 3.27 2.29 2.69 Cluster 4 2.92 2.92 2.11 2.39 2.62 2.65 Cluster 5 2.97 3.19 3.44 2.22 2.95 Cluster 6 3.49 2.25 3.46 4.91 Cluster 7 3.09 2.44 2.11 Cluster 8 3.05 2.43 Cluster 9 3.56 “Goodness of fit” measures Av. distance 3.19 3.07 2.96 2.81 2.72 2.74 2.61 2.83 Min. distance 3.05 2.92 2.69 2.24 2.11 2.22 2.09 1.89 Max. distance 3.33 3.18 3.31 3.06 3.49 3.44 3.46 4.91 Largest outlier 8.012 8.34 7.72 7.35 7.13 7.25 7.54 7.36 Montenegro (2007) Germany (1949) Hungary (2011) Germany (1949) Germany (1949) Germany (1949) Hungary (2011) Montenegro (2007) No. of clusters Average intra-cluster distance 2 3 4 5 6 7 8 9 Cluster 1 3.05 2.92 3.31 3.05 2.81 2.22 2.09 2.58 Cluster 2 3.33 3.11 2.88 2.88 2.37 2.53 2.74 1.89 Cluster 3 3.18 2.69 2.24 2.31 3.27 2.29 2.69 Cluster 4 2.92 2.92 2.11 2.39 2.62 2.65 Cluster 5 2.97 3.19 3.44 2.22 2.95 Cluster 6 3.49 2.25 3.46 4.91 Cluster 7 3.09 2.44 2.11 Cluster 8 3.05 2.43 Cluster 9 3.56 “Goodness of fit” measures Av. distance 3.19 3.07 2.96 2.81 2.72 2.74 2.61 2.83 Min. distance 3.05 2.92 2.69 2.24 2.11 2.22 2.09 1.89 Max. distance 3.33 3.18 3.31 3.06 3.49 3.44 3.46 4.91 Largest outlier 8.012 8.34 7.72 7.35 7.13 7.25 7.54 7.36 Montenegro (2007) Germany (1949) Hungary (2011) Germany (1949) Germany (1949) Germany (1949) Hungary (2011) Montenegro (2007) Note: average, minimum and maximum distances reported in the lower panel are inter-cluster distances. The largest outlier is the distance of the given country to the centroid of the cluster in which it is placed. View Large Following the results in the table toward the right, it is easy to see that the intra-cluster distance across the solutions decreases until the six-cluster solution, to the right of which it increases again. The minimum distance—indicating the coherence of a cluster—also decreases until the six-cluster solution. Likewise, the largest outlier in the six-cluster solution, i.e., the constitution placed the furthest away from other cluster members in 31-dimensional space, is relatively the smallest. Moving beyond a solution with six clusters entails a deterioration in at least one of our “goodness of fit” measures and in particular leaves larger outliers and in all cases at least one comparatively poorly defined cluster, as captured by the average distance. In all cases, we find that three particular constitutions fit any family of constitutions poorly: emergency provisions of the present constitutions of Germany (the 1949 constitution as amended in 1968), Hungary, and Montenegro are structurally different from most other emergency constitutions. The overall intra-cluster coherence on the basis of six clusters is also fairly good in comparison to alternative cluster numbers. This is a final reason for choosing six clusters. Based on the proposed criteria, a solution with six clusters best fit the present data.46Table 5 reports the full cluster membership of this solution for all 351 constitutions in the dataset.47 Table 5. Cluster memberships, preferred six-cluster solution. Cluster Members 1. Domestic security Afghanistan (2004), Albania (1918, 1998), Angola (2010), Antigua and Barbuda (1981), Argentina (1853), Azerbaijan (1995), Barbados (1966), Belarus (1994), Bolivia (1938, 1947, 1961, 1967, 2009), Brazil (1967), Cambodia (1971), Chile (1833, 1980), Congo (1991, 2001), DR Congo (1964), Costa Rica (1917), Cyprus (1960), Ecuador (1869, 1897, 1946, 1967, 1978, 1984, 1998), Eritrea (1997), Estonia (1991), Ethiopia (1994), Ghana (1979, 1991), Guatemala (1945, 1965), Haiti (1889, 1950, 1964, 1983, 1987), Honduras (1965, 1981), Iran (1979), Kenya (2010), Kosovo (2008), Kyrgyzstan (2010), Liberia (1986), Lithuania (1921), Macedonia (1991), Madagascar (1991), Malawi (1994), Maldives (2008), Mexico (1857, 1917), Mongolia (1991), Mozambique (2004), Myanmar (2008), Namibia (1990), Nepal (1990), Nicaragua (1987), Niger (1960), Nigeria (1989, 1999), Panama (1971), Paraguay (1991), Peru (1979), Philippines 1973), Poland (1921, 1935, 1997), Portugal (1976), Samoa (1961), Saudi Arabia (1991), Seychelles (1993), Spain (1931), St. Kitts and Nevis (1983), St. Vincent and the Grenadines (1979), Swaziland (2005), Syria (1953), Tajikistan (1994), Thailand (1968, 1997, 2007), Turkey (1924, 1945, 1961, 1981), Uganda (1995), Uruguay (1918), Uzbekistan (1991), Vanuatu (1980), Venezuela (1999) 2. Middle of the road Afghanistan (1964), Albania (1915), Benin (1964), Bolivia (1851), Bulgaria (1991), Burundi (1981, 1991, 2005), Cameroon (1961), Central African Republic (1981, 2004), China (1913, 1954), Colombia (1886), Comoros (1978), Congo (1963), Cote D’Ivoire (2000), East Timor (2001), Egypt (1971), Estonia (1937), Ethiopia (1955, 1987), Fiji (1970, 1990, 1997), Finland (1999), France (1851), Gabon (1961, 1975), Gambia (1970), Greece (1951), Guatemala (1879), Guinea (1981), Haiti (1843, 1946), Kazakhstan (1995), Kenya (1963), Kiribati (1979), Kuwait (1961), Kyrgyzstan (1993), Latvia (1921), Lithuania (1918, 1938), Madagascar (1998), Maldives (1998), Mali (1974), Mauritania (1991), Mongolia (1960), Morocco (1961), Mozambique (1990), Myanmar (1974), Nauru (1968), Nepal (1948, 1959, 1961), Niger (1996, 1999), Pakistan (1961), Peru (1818), Philippines (1987), Poland (1951, 1991), Qatar (2003), Senegal (2001), Sierra Leone (1978), Solomon Islands (1978), Spain (1978), Togo (1963, 1991), Tuvalu (1986), Venezuela (1961), South Vietnam (1956), Zimbabwe (1979) 3. Serious on rights Albania (1976), Andorra (1993), Bolivia (1816), Bulgaria (1947, 1971), Burkina Faso (1960), Burundi (1974), Central African Republic (1994), Chad (1961), Denmark (1953), Dominican Republic (1947), Equatorial Guinea (1981), Germany (1871), Ghana (1957), Guinea (1990), Honduras (1936), Japan (1889), Jordan (1946), Laos (1991), Madagascar (1975), Mali (1991), Mauritania (1961), Moldova (1994), Mozambique (1975), Nicaragua (1905, 1948, 1950, 1974), Niger (1991), Peru (1867), Somalia (1979), South Africa (1961, 1983), Syria (1950, 1973), Thailand (1949), Togo (1979), Trinidad and Tobago (1961), USA (1789), Venezuela (1904, 1947,1953), Vietnam (1980, 1991), Yugoslavia (1953) 4. Current design Afghanistan (1977, 1990), Albania (1924), Bahamas (1973), Bahrain (2001), Benin (1990), Botswana (1966), Cambodia (1989), Chile (1811), China (1981), Congo (1979), DR Congo (2003), Costa Rica (1949), Djibouti (1991), Egypt (1923, 2011), Equatorial Guinea (1991), Estonia (1920), Fiji (2013), Gambia (1996), Guinea (2010), Guyana (1970), Indonesia (1945), Iraq (2005), Jordan (1951), Kyrgyzstan (2007), Lesotho (1966), Libya (1951), Malawi (1964), Poland (1947), Russia (1993), Rwanda (1961), Senegal (1960, 1963), Seychelles (1979), Soviet Union (1977), Sri Lanka (1978), Sudan (1973, 1998), Surinam (1987), Thailand (1974), Uganda (1967), Ukraine (1996), North Vietnam (1960), Arab Republic of Yemen (1970), Zambia (1964, 1973, 1991), Zimbabwe (1969, 2013) 5. Strong legislature Argentina (1819, 1826), Austria (1934), Brazil (1891, 1946), Chile (1925), Costa Rica (1869), Cuba (1935, 1940), Dominican Republic (1896, 1908, 1914, 1917, 1955, 1961a, 1961b, 1963, 1966, 1994, 2001), El Salvador (1886, 1939, 1950, 1983), France (1946), West Germany (1949), Haiti (1807), Honduras (1880, 1904, 1914, 1957), Hungary (2011), Lithuania (1991), Montenegro (2007), Nicaragua (1893, 1911), Nigeria (1960, 1963), Panama (1904, 1946), Paraguay (1870), Peru (1860, 1910), Portugal (1933), Serbia and Montenegro (2003, 2006), Sierra Leone (1961), South Africa (1993, 1996), Venezuela (1874, 1881, 1891), Yugoslavia (1921) 6. Rights suspenders Armenia (1995), Belize (1981); Bhutan (2008), Bolivia (1878, 1880), Brazil (1937), Burkina Faso (1991), Cameroon (1971), Central African Republic (1964, 1976), Chad (1960, 1996), Comoros (1996), DR Congo (1978, 2005), Cuba (1959), Ecuador (1884, 1906, 1993, 1996, 2008), France (1848, 1958), Ghana (1969), Jamaica (1961), Lesotho (1993), Mauritania (1978), Morocco (1970, 2011), Niger (1989), Oman (1996), Paraguay (1940), Rwanda (2003), Sri Lanka (1971), Yugoslavia (1931, 1974) Cluster Members 1. Domestic security Afghanistan (2004), Albania (1918, 1998), Angola (2010), Antigua and Barbuda (1981), Argentina (1853), Azerbaijan (1995), Barbados (1966), Belarus (1994), Bolivia (1938, 1947, 1961, 1967, 2009), Brazil (1967), Cambodia (1971), Chile (1833, 1980), Congo (1991, 2001), DR Congo (1964), Costa Rica (1917), Cyprus (1960), Ecuador (1869, 1897, 1946, 1967, 1978, 1984, 1998), Eritrea (1997), Estonia (1991), Ethiopia (1994), Ghana (1979, 1991), Guatemala (1945, 1965), Haiti (1889, 1950, 1964, 1983, 1987), Honduras (1965, 1981), Iran (1979), Kenya (2010), Kosovo (2008), Kyrgyzstan (2010), Liberia (1986), Lithuania (1921), Macedonia (1991), Madagascar (1991), Malawi (1994), Maldives (2008), Mexico (1857, 1917), Mongolia (1991), Mozambique (2004), Myanmar (2008), Namibia (1990), Nepal (1990), Nicaragua (1987), Niger (1960), Nigeria (1989, 1999), Panama (1971), Paraguay (1991), Peru (1979), Philippines 1973), Poland (1921, 1935, 1997), Portugal (1976), Samoa (1961), Saudi Arabia (1991), Seychelles (1993), Spain (1931), St. Kitts and Nevis (1983), St. Vincent and the Grenadines (1979), Swaziland (2005), Syria (1953), Tajikistan (1994), Thailand (1968, 1997, 2007), Turkey (1924, 1945, 1961, 1981), Uganda (1995), Uruguay (1918), Uzbekistan (1991), Vanuatu (1980), Venezuela (1999) 2. Middle of the road Afghanistan (1964), Albania (1915), Benin (1964), Bolivia (1851), Bulgaria (1991), Burundi (1981, 1991, 2005), Cameroon (1961), Central African Republic (1981, 2004), China (1913, 1954), Colombia (1886), Comoros (1978), Congo (1963), Cote D’Ivoire (2000), East Timor (2001), Egypt (1971), Estonia (1937), Ethiopia (1955, 1987), Fiji (1970, 1990, 1997), Finland (1999), France (1851), Gabon (1961, 1975), Gambia (1970), Greece (1951), Guatemala (1879), Guinea (1981), Haiti (1843, 1946), Kazakhstan (1995), Kenya (1963), Kiribati (1979), Kuwait (1961), Kyrgyzstan (1993), Latvia (1921), Lithuania (1918, 1938), Madagascar (1998), Maldives (1998), Mali (1974), Mauritania (1991), Mongolia (1960), Morocco (1961), Mozambique (1990), Myanmar (1974), Nauru (1968), Nepal (1948, 1959, 1961), Niger (1996, 1999), Pakistan (1961), Peru (1818), Philippines (1987), Poland (1951, 1991), Qatar (2003), Senegal (2001), Sierra Leone (1978), Solomon Islands (1978), Spain (1978), Togo (1963, 1991), Tuvalu (1986), Venezuela (1961), South Vietnam (1956), Zimbabwe (1979) 3. Serious on rights Albania (1976), Andorra (1993), Bolivia (1816), Bulgaria (1947, 1971), Burkina Faso (1960), Burundi (1974), Central African Republic (1994), Chad (1961), Denmark (1953), Dominican Republic (1947), Equatorial Guinea (1981), Germany (1871), Ghana (1957), Guinea (1990), Honduras (1936), Japan (1889), Jordan (1946), Laos (1991), Madagascar (1975), Mali (1991), Mauritania (1961), Moldova (1994), Mozambique (1975), Nicaragua (1905, 1948, 1950, 1974), Niger (1991), Peru (1867), Somalia (1979), South Africa (1961, 1983), Syria (1950, 1973), Thailand (1949), Togo (1979), Trinidad and Tobago (1961), USA (1789), Venezuela (1904, 1947,1953), Vietnam (1980, 1991), Yugoslavia (1953) 4. Current design Afghanistan (1977, 1990), Albania (1924), Bahamas (1973), Bahrain (2001), Benin (1990), Botswana (1966), Cambodia (1989), Chile (1811), China (1981), Congo (1979), DR Congo (2003), Costa Rica (1949), Djibouti (1991), Egypt (1923, 2011), Equatorial Guinea (1991), Estonia (1920), Fiji (2013), Gambia (1996), Guinea (2010), Guyana (1970), Indonesia (1945), Iraq (2005), Jordan (1951), Kyrgyzstan (2007), Lesotho (1966), Libya (1951), Malawi (1964), Poland (1947), Russia (1993), Rwanda (1961), Senegal (1960, 1963), Seychelles (1979), Soviet Union (1977), Sri Lanka (1978), Sudan (1973, 1998), Surinam (1987), Thailand (1974), Uganda (1967), Ukraine (1996), North Vietnam (1960), Arab Republic of Yemen (1970), Zambia (1964, 1973, 1991), Zimbabwe (1969, 2013) 5. Strong legislature Argentina (1819, 1826), Austria (1934), Brazil (1891, 1946), Chile (1925), Costa Rica (1869), Cuba (1935, 1940), Dominican Republic (1896, 1908, 1914, 1917, 1955, 1961a, 1961b, 1963, 1966, 1994, 2001), El Salvador (1886, 1939, 1950, 1983), France (1946), West Germany (1949), Haiti (1807), Honduras (1880, 1904, 1914, 1957), Hungary (2011), Lithuania (1991), Montenegro (2007), Nicaragua (1893, 1911), Nigeria (1960, 1963), Panama (1904, 1946), Paraguay (1870), Peru (1860, 1910), Portugal (1933), Serbia and Montenegro (2003, 2006), Sierra Leone (1961), South Africa (1993, 1996), Venezuela (1874, 1881, 1891), Yugoslavia (1921) 6. Rights suspenders Armenia (1995), Belize (1981); Bhutan (2008), Bolivia (1878, 1880), Brazil (1937), Burkina Faso (1991), Cameroon (1971), Central African Republic (1964, 1976), Chad (1960, 1996), Comoros (1996), DR Congo (1978, 2005), Cuba (1959), Ecuador (1884, 1906, 1993, 1996, 2008), France (1848, 1958), Ghana (1969), Jamaica (1961), Lesotho (1993), Mauritania (1978), Morocco (1970, 2011), Niger (1989), Oman (1996), Paraguay (1940), Rwanda (2003), Sri Lanka (1971), Yugoslavia (1931, 1974) View Large Table 5. Cluster memberships, preferred six-cluster solution. Cluster Members 1. Domestic security Afghanistan (2004), Albania (1918, 1998), Angola (2010), Antigua and Barbuda (1981), Argentina (1853), Azerbaijan (1995), Barbados (1966), Belarus (1994), Bolivia (1938, 1947, 1961, 1967, 2009), Brazil (1967), Cambodia (1971), Chile (1833, 1980), Congo (1991, 2001), DR Congo (1964), Costa Rica (1917), Cyprus (1960), Ecuador (1869, 1897, 1946, 1967, 1978, 1984, 1998), Eritrea (1997), Estonia (1991), Ethiopia (1994), Ghana (1979, 1991), Guatemala (1945, 1965), Haiti (1889, 1950, 1964, 1983, 1987), Honduras (1965, 1981), Iran (1979), Kenya (2010), Kosovo (2008), Kyrgyzstan (2010), Liberia (1986), Lithuania (1921), Macedonia (1991), Madagascar (1991), Malawi (1994), Maldives (2008), Mexico (1857, 1917), Mongolia (1991), Mozambique (2004), Myanmar (2008), Namibia (1990), Nepal (1990), Nicaragua (1987), Niger (1960), Nigeria (1989, 1999), Panama (1971), Paraguay (1991), Peru (1979), Philippines 1973), Poland (1921, 1935, 1997), Portugal (1976), Samoa (1961), Saudi Arabia (1991), Seychelles (1993), Spain (1931), St. Kitts and Nevis (1983), St. Vincent and the Grenadines (1979), Swaziland (2005), Syria (1953), Tajikistan (1994), Thailand (1968, 1997, 2007), Turkey (1924, 1945, 1961, 1981), Uganda (1995), Uruguay (1918), Uzbekistan (1991), Vanuatu (1980), Venezuela (1999) 2. Middle of the road Afghanistan (1964), Albania (1915), Benin (1964), Bolivia (1851), Bulgaria (1991), Burundi (1981, 1991, 2005), Cameroon (1961), Central African Republic (1981, 2004), China (1913, 1954), Colombia (1886), Comoros (1978), Congo (1963), Cote D’Ivoire (2000), East Timor (2001), Egypt (1971), Estonia (1937), Ethiopia (1955, 1987), Fiji (1970, 1990, 1997), Finland (1999), France (1851), Gabon (1961, 1975), Gambia (1970), Greece (1951), Guatemala (1879), Guinea (1981), Haiti (1843, 1946), Kazakhstan (1995), Kenya (1963), Kiribati (1979), Kuwait (1961), Kyrgyzstan (1993), Latvia (1921), Lithuania (1918, 1938), Madagascar (1998), Maldives (1998), Mali (1974), Mauritania (1991), Mongolia (1960), Morocco (1961), Mozambique (1990), Myanmar (1974), Nauru (1968), Nepal (1948, 1959, 1961), Niger (1996, 1999), Pakistan (1961), Peru (1818), Philippines (1987), Poland (1951, 1991), Qatar (2003), Senegal (2001), Sierra Leone (1978), Solomon Islands (1978), Spain (1978), Togo (1963, 1991), Tuvalu (1986), Venezuela (1961), South Vietnam (1956), Zimbabwe (1979) 3. Serious on rights Albania (1976), Andorra (1993), Bolivia (1816), Bulgaria (1947, 1971), Burkina Faso (1960), Burundi (1974), Central African Republic (1994), Chad (1961), Denmark (1953), Dominican Republic (1947), Equatorial Guinea (1981), Germany (1871), Ghana (1957), Guinea (1990), Honduras (1936), Japan (1889), Jordan (1946), Laos (1991), Madagascar (1975), Mali (1991), Mauritania (1961), Moldova (1994), Mozambique (1975), Nicaragua (1905, 1948, 1950, 1974), Niger (1991), Peru (1867), Somalia (1979), South Africa (1961, 1983), Syria (1950, 1973), Thailand (1949), Togo (1979), Trinidad and Tobago (1961), USA (1789), Venezuela (1904, 1947,1953), Vietnam (1980, 1991), Yugoslavia (1953) 4. Current design Afghanistan (1977, 1990), Albania (1924), Bahamas (1973), Bahrain (2001), Benin (1990), Botswana (1966), Cambodia (1989), Chile (1811), China (1981), Congo (1979), DR Congo (2003), Costa Rica (1949), Djibouti (1991), Egypt (1923, 2011), Equatorial Guinea (1991), Estonia (1920), Fiji (2013), Gambia (1996), Guinea (2010), Guyana (1970), Indonesia (1945), Iraq (2005), Jordan (1951), Kyrgyzstan (2007), Lesotho (1966), Libya (1951), Malawi (1964), Poland (1947), Russia (1993), Rwanda (1961), Senegal (1960, 1963), Seychelles (1979), Soviet Union (1977), Sri Lanka (1978), Sudan (1973, 1998), Surinam (1987), Thailand (1974), Uganda (1967), Ukraine (1996), North Vietnam (1960), Arab Republic of Yemen (1970), Zambia (1964, 1973, 1991), Zimbabwe (1969, 2013) 5. Strong legislature Argentina (1819, 1826), Austria (1934), Brazil (1891, 1946), Chile (1925), Costa Rica (1869), Cuba (1935, 1940), Dominican Republic (1896, 1908, 1914, 1917, 1955, 1961a, 1961b, 1963, 1966, 1994, 2001), El Salvador (1886, 1939, 1950, 1983), France (1946), West Germany (1949), Haiti (1807), Honduras (1880, 1904, 1914, 1957), Hungary (2011), Lithuania (1991), Montenegro (2007), Nicaragua (1893, 1911), Nigeria (1960, 1963), Panama (1904, 1946), Paraguay (1870), Peru (1860, 1910), Portugal (1933), Serbia and Montenegro (2003, 2006), Sierra Leone (1961), South Africa (1993, 1996), Venezuela (1874, 1881, 1891), Yugoslavia (1921) 6. Rights suspenders Armenia (1995), Belize (1981); Bhutan (2008), Bolivia (1878, 1880), Brazil (1937), Burkina Faso (1991), Cameroon (1971), Central African Republic (1964, 1976), Chad (1960, 1996), Comoros (1996), DR Congo (1978, 2005), Cuba (1959), Ecuador (1884, 1906, 1993, 1996, 2008), France (1848, 1958), Ghana (1969), Jamaica (1961), Lesotho (1993), Mauritania (1978), Morocco (1970, 2011), Niger (1989), Oman (1996), Paraguay (1940), Rwanda (2003), Sri Lanka (1971), Yugoslavia (1931, 1974) Cluster Members 1. Domestic security Afghanistan (2004), Albania (1918, 1998), Angola (2010), Antigua and Barbuda (1981), Argentina (1853), Azerbaijan (1995), Barbados (1966), Belarus (1994), Bolivia (1938, 1947, 1961, 1967, 2009), Brazil (1967), Cambodia (1971), Chile (1833, 1980), Congo (1991, 2001), DR Congo (1964), Costa Rica (1917), Cyprus (1960), Ecuador (1869, 1897, 1946, 1967, 1978, 1984, 1998), Eritrea (1997), Estonia (1991), Ethiopia (1994), Ghana (1979, 1991), Guatemala (1945, 1965), Haiti (1889, 1950, 1964, 1983, 1987), Honduras (1965, 1981), Iran (1979), Kenya (2010), Kosovo (2008), Kyrgyzstan (2010), Liberia (1986), Lithuania (1921), Macedonia (1991), Madagascar (1991), Malawi (1994), Maldives (2008), Mexico (1857, 1917), Mongolia (1991), Mozambique (2004), Myanmar (2008), Namibia (1990), Nepal (1990), Nicaragua (1987), Niger (1960), Nigeria (1989, 1999), Panama (1971), Paraguay (1991), Peru (1979), Philippines 1973), Poland (1921, 1935, 1997), Portugal (1976), Samoa (1961), Saudi Arabia (1991), Seychelles (1993), Spain (1931), St. Kitts and Nevis (1983), St. Vincent and the Grenadines (1979), Swaziland (2005), Syria (1953), Tajikistan (1994), Thailand (1968, 1997, 2007), Turkey (1924, 1945, 1961, 1981), Uganda (1995), Uruguay (1918), Uzbekistan (1991), Vanuatu (1980), Venezuela (1999) 2. Middle of the road Afghanistan (1964), Albania (1915), Benin (1964), Bolivia (1851), Bulgaria (1991), Burundi (1981, 1991, 2005), Cameroon (1961), Central African Republic (1981, 2004), China (1913, 1954), Colombia (1886), Comoros (1978), Congo (1963), Cote D’Ivoire (2000), East Timor (2001), Egypt (1971), Estonia (1937), Ethiopia (1955, 1987), Fiji (1970, 1990, 1997), Finland (1999), France (1851), Gabon (1961, 1975), Gambia (1970), Greece (1951), Guatemala (1879), Guinea (1981), Haiti (1843, 1946), Kazakhstan (1995), Kenya (1963), Kiribati (1979), Kuwait (1961), Kyrgyzstan (1993), Latvia (1921), Lithuania (1918, 1938), Madagascar (1998), Maldives (1998), Mali (1974), Mauritania (1991), Mongolia (1960), Morocco (1961), Mozambique (1990), Myanmar (1974), Nauru (1968), Nepal (1948, 1959, 1961), Niger (1996, 1999), Pakistan (1961), Peru (1818), Philippines (1987), Poland (1951, 1991), Qatar (2003), Senegal (2001), Sierra Leone (1978), Solomon Islands (1978), Spain (1978), Togo (1963, 1991), Tuvalu (1986), Venezuela (1961), South Vietnam (1956), Zimbabwe (1979) 3. Serious on rights Albania (1976), Andorra (1993), Bolivia (1816), Bulgaria (1947, 1971), Burkina Faso (1960), Burundi (1974), Central African Republic (1994), Chad (1961), Denmark (1953), Dominican Republic (1947), Equatorial Guinea (1981), Germany (1871), Ghana (1957), Guinea (1990), Honduras (1936), Japan (1889), Jordan (1946), Laos (1991), Madagascar (1975), Mali (1991), Mauritania (1961), Moldova (1994), Mozambique (1975), Nicaragua (1905, 1948, 1950, 1974), Niger (1991), Peru (1867), Somalia (1979), South Africa (1961, 1983), Syria (1950, 1973), Thailand (1949), Togo (1979), Trinidad and Tobago (1961), USA (1789), Venezuela (1904, 1947,1953), Vietnam (1980, 1991), Yugoslavia (1953) 4. Current design Afghanistan (1977, 1990), Albania (1924), Bahamas (1973), Bahrain (2001), Benin (1990), Botswana (1966), Cambodia (1989), Chile (1811), China (1981), Congo (1979), DR Congo (2003), Costa Rica (1949), Djibouti (1991), Egypt (1923, 2011), Equatorial Guinea (1991), Estonia (1920), Fiji (2013), Gambia (1996), Guinea (2010), Guyana (1970), Indonesia (1945), Iraq (2005), Jordan (1951), Kyrgyzstan (2007), Lesotho (1966), Libya (1951), Malawi (1964), Poland (1947), Russia (1993), Rwanda (1961), Senegal (1960, 1963), Seychelles (1979), Soviet Union (1977), Sri Lanka (1978), Sudan (1973, 1998), Surinam (1987), Thailand (1974), Uganda (1967), Ukraine (1996), North Vietnam (1960), Arab Republic of Yemen (1970), Zambia (1964, 1973, 1991), Zimbabwe (1969, 2013) 5. Strong legislature Argentina (1819, 1826), Austria (1934), Brazil (1891, 1946), Chile (1925), Costa Rica (1869), Cuba (1935, 1940), Dominican Republic (1896, 1908, 1914, 1917, 1955, 1961a, 1961b, 1963, 1966, 1994, 2001), El Salvador (1886, 1939, 1950, 1983), France (1946), West Germany (1949), Haiti (1807), Honduras (1880, 1904, 1914, 1957), Hungary (2011), Lithuania (1991), Montenegro (2007), Nicaragua (1893, 1911), Nigeria (1960, 1963), Panama (1904, 1946), Paraguay (1870), Peru (1860, 1910), Portugal (1933), Serbia and Montenegro (2003, 2006), Sierra Leone (1961), South Africa (1993, 1996), Venezuela (1874, 1881, 1891), Yugoslavia (1921) 6. Rights suspenders Armenia (1995), Belize (1981); Bhutan (2008), Bolivia (1878, 1880), Brazil (1937), Burkina Faso (1991), Cameroon (1971), Central African Republic (1964, 1976), Chad (1960, 1996), Comoros (1996), DR Congo (1978, 2005), Cuba (1959), Ecuador (1884, 1906, 1993, 1996, 2008), France (1848, 1958), Ghana (1969), Jamaica (1961), Lesotho (1993), Mauritania (1978), Morocco (1970, 2011), Niger (1989), Oman (1996), Paraguay (1940), Rwanda (2003), Sri Lanka (1971), Yugoslavia (1931, 1974) View Large 4.2. The cluster solution Although cluster analysis can at times yield solutions that are difficult to interpret, the present six-cluster solution in the table indicates that the clusters are all defined by traits common to their specific member constitutions. Cluster 1 thus appears to be defined by a concern for domestic security,48 cluster 3 by having particularly well-defined rights protection, while cluster 4 contains many relatively recent constitutions and thus represents what may be termed current design. Cluster 2, on the other hand, is characterized by emergency constitutions that on most counts represent an average constitutional design. Finally, the average within-country distances indicate that clusters 5 and 6 may be less clearly defined than the four first clusters (based on information contained in Table 4). We nevertheless find that constitutions allocated to clusters 5 and 6 share particular identifiable features. Overall, it turns out that most of the differences in the various types of emergency constitutions can be traced back to four variables, namely, (i) legislative declaration, i.e., the role of the legislature in declaring a state of emergency; (ii) the specific approval powers; (iii) the specific conditions that allow declaring a state of emergency; and (4) which rights, if any, can be suspended during emergencies. We therefore interpret these final clusters as defined by having strong legislatures and easy rights suspension, respectively. We proceed with a number of noteworthy observations summarized in Table 6. Table 6. Cluster characteristics. 1 2 3 4 5 6 Domesticsecurity Middle of the road Serious on rights Current design Strong legislature Easy rights suspenders Centroid Guatemala 1945 Congo 1963 Bulgaria 1947 Zimbabwe 1969 Paraguay 1870 Ecuador 1906 Av. distance 2.81 2.37 2.31 2.11 3.19 3.49 Members 94 73 45 50 53 36 INEP 0.36 0.33 0.26 0.34 0.23 0.49 Declaration power 0.94 0.92 0.64 0.95 0.07 Legislative declaration 0.94 Approval power 0.13 Legislative approval 0.29 0.24 Vague approval prov. 0.21 0.38 0.29 Other approval Conditions 0.50 Internal security 0.23 Nat. disasters not mentioned 0.15 War, nat. disasters not mentioned 0.13 Vague conditions 0.39 0.39 Dissolution power 0.07 0.08 0.04 0.20 0.06 0.31 Rights suspension 0.21 Rights suspension 0.07 0.06 0.12 .23 Rights suspension 0.56 Some rights suspension Expropriation and censorship 0.32 0.40 Censorship 0.40 Censorship 0.44 0.27 0.44 Censorship Age 45 51 62 42 84 48 Length 18,706 13,138 8,727 15,800 13,466 15,226 Legal origin 83% civil 79% civil 67% civil 60% civil 91% civil 81% civil Democracy 37% 25% 20% 14% 38% 33% PolConIII 0.10 0.098 0.063 0.065 0.12 0.19 GDP at birth 2,905 2,565 2,062 2,443 2,361 2,268 Still in use 18 6 6 22 8 13 Post-Cold War 39% 29% 18% 34% 17% 36% 1 2 3 4 5 6 Domesticsecurity Middle of the road Serious on rights Current design Strong legislature Easy rights suspenders Centroid Guatemala 1945 Congo 1963 Bulgaria 1947 Zimbabwe 1969 Paraguay 1870 Ecuador 1906 Av. distance 2.81 2.37 2.31 2.11 3.19 3.49 Members 94 73 45 50 53 36 INEP 0.36 0.33 0.26 0.34 0.23 0.49 Declaration power 0.94 0.92 0.64 0.95 0.07 Legislative declaration 0.94 Approval power 0.13 Legislative approval 0.29 0.24 Vague approval prov. 0.21 0.38 0.29 Other approval Conditions 0.50 Internal security 0.23 Nat. disasters not mentioned 0.15 War, nat. disasters not mentioned 0.13 Vague conditions 0.39 0.39 Dissolution power 0.07 0.08 0.04 0.20 0.06 0.31 Rights suspension 0.21 Rights suspension 0.07 0.06 0.12 .23 Rights suspension 0.56 Some rights suspension Expropriation and censorship 0.32 0.40 Censorship 0.40 Censorship 0.44 0.27 0.44 Censorship Age 45 51 62 42 84 48 Length 18,706 13,138 8,727 15,800 13,466 15,226 Legal origin 83% civil 79% civil 67% civil 60% civil 91% civil 81% civil Democracy 37% 25% 20% 14% 38% 33% PolConIII 0.10 0.098 0.063 0.065 0.12 0.19 GDP at birth 2,905 2,565 2,062 2,443 2,361 2,268 Still in use 18 6 6 22 8 13 Post-Cold War 39% 29% 18% 34% 17% 36% View Large Table 6. Cluster characteristics. 1 2 3 4 5 6 Domesticsecurity Middle of the road Serious on rights Current design Strong legislature Easy rights suspenders Centroid Guatemala 1945 Congo 1963 Bulgaria 1947 Zimbabwe 1969 Paraguay 1870 Ecuador 1906 Av. distance 2.81 2.37 2.31 2.11 3.19 3.49 Members 94 73 45 50 53 36 INEP 0.36 0.33 0.26 0.34 0.23 0.49 Declaration power 0.94 0.92 0.64 0.95 0.07 Legislative declaration 0.94 Approval power 0.13 Legislative approval 0.29 0.24 Vague approval prov. 0.21 0.38 0.29 Other approval Conditions 0.50 Internal security 0.23 Nat. disasters not mentioned 0.15 War, nat. disasters not mentioned 0.13 Vague conditions 0.39 0.39 Dissolution power 0.07 0.08 0.04 0.20 0.06 0.31 Rights suspension 0.21 Rights suspension 0.07 0.06 0.12 .23 Rights suspension 0.56 Some rights suspension Expropriation and censorship 0.32 0.40 Censorship 0.40 Censorship 0.44 0.27 0.44 Censorship Age 45 51 62 42 84 48 Length 18,706 13,138 8,727 15,800 13,466 15,226 Legal origin 83% civil 79% civil 67% civil 60% civil 91% civil 81% civil Democracy 37% 25% 20% 14% 38% 33% PolConIII 0.10 0.098 0.063 0.065 0.12 0.19 GDP at birth 2,905 2,565 2,062 2,443 2,361 2,268 Still in use 18 6 6 22 8 13 Post-Cold War 39% 29% 18% 34% 17% 36% 1 2 3 4 5 6 Domesticsecurity Middle of the road Serious on rights Current design Strong legislature Easy rights suspenders Centroid Guatemala 1945 Congo 1963 Bulgaria 1947 Zimbabwe 1969 Paraguay 1870 Ecuador 1906 Av. distance 2.81 2.37 2.31 2.11 3.19 3.49 Members 94 73 45 50 53 36 INEP 0.36 0.33 0.26 0.34 0.23 0.49 Declaration power 0.94 0.92 0.64 0.95 0.07 Legislative declaration 0.94 Approval power 0.13 Legislative approval 0.29 0.24 Vague approval prov. 0.21 0.38 0.29 Other approval Conditions 0.50 Internal security 0.23 Nat. disasters not mentioned 0.15 War, nat. disasters not mentioned 0.13 Vague conditions 0.39 0.39 Dissolution power 0.07 0.08 0.04 0.20 0.06 0.31 Rights suspension 0.21 Rights suspension 0.07 0.06 0.12 .23 Rights suspension 0.56 Some rights suspension Expropriation and censorship 0.32 0.40 Censorship 0.40 Censorship 0.44 0.27 0.44 Censorship Age 45 51 62 42 84 48 Length 18,706 13,138 8,727 15,800 13,466 15,226 Legal origin 83% civil 79% civil 67% civil 60% civil 91% civil 81% civil Democracy 37% 25% 20% 14% 38% 33% PolConIII 0.10 0.098 0.063 0.065 0.12 0.19 GDP at birth 2,905 2,565 2,062 2,443 2,361 2,268 Still in use 18 6 6 22 8 13 Post-Cold War 39% 29% 18% 34% 17% 36% View Large Let us begin with those constitutions not grouped in any of the six clusters because they do not have emergency constitutions. We identify fifty-eight constitutions from twenty-one countries as such constitutions. On average, these constitutions are older, shorter, and less likely to be located in a civil law system and all those constitutions containing emergency provisions. Of the twelve constitutions without emergency provisions that remain in use today, five belong to systems coded as democratic according to Cheibub et al.49 These democracies are in no way clearly different from democracies in which the constitutions include emergency provisions. The second noteworthy observation is that the mean INEP for clusters 1, 2 and 4 is fairly similar at around 0.35 whereas it is substantially lower in clusters 3 and 5 (around 0.25) and much larger in clusters 6 at 0.49. However, as is evident in Table 6, the regular constitutions in countries in cluster 6 also have the strongest veto player institutions, as measured by Henisz’s50 (2010) PolConIII index. As such, regardless of the mix of characteristics, cluster 6 seems to be characterized by a substantially larger difference between the separation of powers in ordinary versus emergency times than other clusters. We also note that constitutions in cluster 1 are, in general, substantially longer and thus more detailed than in other clusters whereas constitutions in cluster 3 are shorter. Conversely, comparing legal origins and real PPP (purchasing power parity)-adjusted GDP (gross domestic product) per capita at the time when the constitution was introduced, which we take from the Maddison database,51 we find no discernible differences in legal origins or GDP, while constitutions in cluster 3 tend to be older and very few constitutions in cluster 4 were written when the country was democratic. This is a rather unexpected finding given the history of emergency powers described in Section 2 of this article. We capture the particular mix of powers in an alternative way in the multinomial logistic regression reported in Table 7, which clearly shows that—relative to the baseline of cluster 1—emergency provisions in clusters 3 and 5 are substantially and significantly more restricted than in the remaining clusters. Clusters 1 and 4 stand out with weaker approval powers than the rest although it is also clear that cluster 1 is characterized by having legislative approval while approval in cluster 6 is left to “other,” meaning non-political actors. Regarding the right to suspend basic rights, this is strongly circumscribed in cluster 3 and broadly allowed in constitutions in clusters 2 and 6. Finally, clusters 3, 4, and 6 are characterized by allowing censorship and expropriation without ordinary compensation during emergencies. Table 7. Determinants of cluster placement, constitutional characteristics Cluster 2 3 4 5 6 Declaration power −0.506 (0.796) −2.682*** (0.789) 0.402 (0.992) −7.835*** (1.159) −0.156 (0.979) Approval power 2.329*** (0.745) 2.156** (0.912) 1.329 (0.880) 5.632*** (1.326) 2.515*** (0.853) Conditions −6.874*** (1.169) −10.903*** (1.716) −12.848*** (1.774) 0.528 (1.682) −2.999*** (1.131) Dissolution power −0.427 (0.655) −1.668* (0.993) 0.508 (0.666) −1.366 (1.019) 0.929 (0.643) Rights suspension 2.958*** (1.058) −4.279*** (1.545) −1.099 (1.045) −2.547 (1.617) 3.267*** (0.745) Expropriation and censorship 1.633* (0.863) 2.292** (1.115) 2.697** (1.118) 2.234* (1.263) 1.964* (1.085) Observations 351 LR chi squared 490.76 Pseudo R squared 0.402 Cluster 2 3 4 5 6 Declaration power −0.506 (0.796) −2.682*** (0.789) 0.402 (0.992) −7.835*** (1.159) −0.156 (0.979) Approval power 2.329*** (0.745) 2.156** (0.912) 1.329 (0.880) 5.632*** (1.326) 2.515*** (0.853) Conditions −6.874*** (1.169) −10.903*** (1.716) −12.848*** (1.774) 0.528 (1.682) −2.999*** (1.131) Dissolution power −0.427 (0.655) −1.668* (0.993) 0.508 (0.666) −1.366 (1.019) 0.929 (0.643) Rights suspension 2.958*** (1.058) −4.279*** (1.545) −1.099 (1.045) −2.547 (1.617) 3.267*** (0.745) Expropriation and censorship 1.633* (0.863) 2.292** (1.115) 2.697** (1.118) 2.234* (1.263) 1.964* (1.085) Observations 351 LR chi squared 490.76 Pseudo R squared 0.402 *Denotes significance at p < .10. ** Denotes significance at p < .05. *** Denotes significance at p < .01. View Large Table 7. Determinants of cluster placement, constitutional characteristics Cluster 2 3 4 5 6 Declaration power −0.506 (0.796) −2.682*** (0.789) 0.402 (0.992) −7.835*** (1.159) −0.156 (0.979) Approval power 2.329*** (0.745) 2.156** (0.912) 1.329 (0.880) 5.632*** (1.326) 2.515*** (0.853) Conditions −6.874*** (1.169) −10.903*** (1.716) −12.848*** (1.774) 0.528 (1.682) −2.999*** (1.131) Dissolution power −0.427 (0.655) −1.668* (0.993) 0.508 (0.666) −1.366 (1.019) 0.929 (0.643) Rights suspension 2.958*** (1.058) −4.279*** (1.545) −1.099 (1.045) −2.547 (1.617) 3.267*** (0.745) Expropriation and censorship 1.633* (0.863) 2.292** (1.115) 2.697** (1.118) 2.234* (1.263) 1.964* (1.085) Observations 351 LR chi squared 490.76 Pseudo R squared 0.402 Cluster 2 3 4 5 6 Declaration power −0.506 (0.796) −2.682*** (0.789) 0.402 (0.992) −7.835*** (1.159) −0.156 (0.979) Approval power 2.329*** (0.745) 2.156** (0.912) 1.329 (0.880) 5.632*** (1.326) 2.515*** (0.853) Conditions −6.874*** (1.169) −10.903*** (1.716) −12.848*** (1.774) 0.528 (1.682) −2.999*** (1.131) Dissolution power −0.427 (0.655) −1.668* (0.993) 0.508 (0.666) −1.366 (1.019) 0.929 (0.643) Rights suspension 2.958*** (1.058) −4.279*** (1.545) −1.099 (1.045) −2.547 (1.617) 3.267*** (0.745) Expropriation and censorship 1.633* (0.863) 2.292** (1.115) 2.697** (1.118) 2.234* (1.263) 1.964* (1.085) Observations 351 LR chi squared 490.76 Pseudo R squared 0.402 *Denotes significance at p < .10. ** Denotes significance at p < .05. *** Denotes significance at p < .01. View Large As such, a simple and preliminary cluster analysis reveals that one can to some extent identify “ideal” types of emergency constitution design. As such, although we do not want to overstate the degree of familiarity within each cluster, the standard claim that each emergency constitution is a unique document that does not lend itself to easy comparison seems at least questionable. To sum up, we observe that emergency constitutions can be separated into six clusters. These clusters differ not only in the make-up of the emergency provisions and the specific degree powers given to the executive, but also in their likelihood of declaring a state of emergency if an actual emergency arises. It is with this background that we now turn to our theoretical considerations and an intuitive test of whether the distribution of countries within the cluster families is stable over time. 4.3. How stable is the cluster assignment for countries over time? As a final element in describing the families of emergency constitutions, we provide information on how stable a country’s placement in a particular cluster or constitutional family is. This exercise requires that we can observe constitutional changes in the emergency provisions within countries. The data include 284 constitutions from eighty-one countries on which we have sufficient information.52 The data first of all shows that eight countries represented by more than one emergency constitution in the data have not changed clusters and therefore constitutional families: El Salvador (four emergency constitutions), Fiji (three), Gabon (two), Mexico (two), Russia and the Soviet Union (two), Sudan (two), Turkey (four), and Zambia (three). Their constitutional choices of emergency provisions remain quite constant over the years due to either political tradition, geographical circumstance, or mere coincidence. Conversely, of the fifty-three countries with at least three constitutions in the dataset, twenty-three had emergency constitutions represented in three different families and Albania, Bolivia, Niger, Peru, and Venezuela have all had emergency constitutions belonging to four different families. However, many of these constitutional changes of course occur between families that are not particularly different. We therefore further note that comparing the Euclidean distance between cluster centroids, the largest relative changes occur between clusters 1 and 3 and clusters 4 and 5. Conversely, clusters 1 and 2 and clusters 2 and 6 are more similar than other cluster pairs. One should therefore not overestimate the importance of emergency constitutions changing constitutional families. The countries with the relatively largest changes are: Albania in clusters 1 (1918 and 1998) and 3 (1976), Bolivia in clusters 3 (1816) and 1 (all twentieth-century constitutions); Chile in clusters 4 (1811) and 5 (1925); Costa Rica in clusters 4 (1949) and 5 (1869); Ghana in clusters 3 (1957) and 1 (subsequent constitutions); Honduras in clusters 3 (1936) and 1 (1965 and 1981); Madagascar in clusters 3 (1975) and 1 (1991); and Mozambique in clusters 1 (2004) and 3 (1975). Nicaragua, whose 1987 constitution in cluster 1 was also substantially different from its other emergency constitutions (all in cluster 3); Niger in clusters 1 (1960) and 3 (1991); Peru in clusters 3 (1867) and 1 (1979); Thailand whose 1949 constitution in cluster 3 was substantially different from its subsequent emergency constitutions; and Venezuela in which Hugo Chavez’s 1999 socialist constitution in cluster 1 had very different emergency provisions than previous constitutions (all in cluster 3). We do not want to delve deeper into the reasons for these changes except to note two regularities. First, in the cases of the 1991 Bulgarian and the 2003 Serbian emergency constitutions, it is evident that they both represented a constitutional return to the constitutional family that their constitutions prior to communism had belonged to. The second regularity is that Latin American countries are substantially more likely to have experienced larger changes to their emergency constitutions. While these countries have also had the most constitutional changes, 47 of the 100 emergency constitutions in the twenty Latin American countries with recorded changes have occurred in countries that have changed constitutional family. Eleven of the twenty countries have even had constitutions in at least three clusters. In sum, identifying a limited number of “typical” emergency constitutions seems very well possible via cluster analysis. Generally, the content of many constitutions changes only marginally even after regime changes.53 This is different with regard to emergency constitutions. Here, shifts between clusters are not at all uncommon. 5. Conclusions and outlook Given that nine out of ten constitutions contain emergency provisions, it is amazing how little we know about them. This article tries to reduce our ignorance by doing two things. First, it contains an Index of Emergency Powers, which measures the degree of discretionary power constitutionally allocated to the executive during emergencies. On average, the overall degree in the separation of powers outlined by constitutions has increased over the last decades. Relying on the INEP, we find that this is emphatically not true for emergency constitutions. Quite to the contrary, we observe a clear long-run trend to allocate more, rather than less, powers to the executive during times of emergency. Our data suggests that this trend has been a consequence of the relatively weak separation of powers in the constitutions of newly independent countries since the late 1950s, and the particular features of constitutions in the countries that became de facto independent after the collapse of the communist bloc in the early 1990s. Second, this article deals with the presumption that emergency constitutions are unique and do not lend themselves to straightforward comparisons with other emergency constitutions. Drawing on thirty-one different variables and employing cluster analysis, we are able to identify six clusters in which we group 351 different constitutions for which we have sufficient information. These clusters each define what can be thought of as a “family” of emergency constitutions that are sufficiently similar to be characterized by very comparable constitutional choices. The main features that separate these families of emergency constitutions are choices regarding the role of the legislature in declaring a state of emergency, the specific approval powers, the constitutional conditions that allow declaring a state of emergency, and which basic individual rights, if any, can be suspended during emergencies. This article is interested in making emergency constitutions comparable. The indicator of emergency powers (the INEP) as well as the clusters here introduced promise to be relevant for a number of highly topical follow-up questions. First, what are the factors that determine what kind of emergency constitution a country adopts, given that it adopts one? Is being prone to natural disasters as prominent a factor as one might expect or are other, less salient ones, even more important? Bjørnskov and Voigt54 is a first attempt to answer these questions. Second, one might ask what the determinants of actually declaring a state of emergency are. Can one show that emergency constitutions entailing more separation of powers are used less frequently? What about the use of states of emergency in countries without explicit emergency provisions? Bjørnskov and Voigt55 deal with these questions. Third, what are the effects of emergency constitutions? Given that we have identified six different components of emergency powers, can one identify one as particularly apt in re-establishing the status quo ante? Or, in minimizing the number of dead after a natural disaster has occurred? Or as inadequate in re-establishing ex ante levels of civil rights? How do countries without an explicit emergency constitution fare in comparison? Fourth, emergency constitutions are but one tool that can serve the interests of power-maximizing politicians. Other tools include, but are not restricted to, the frequent use of executive decrees, the creation of special courts, and the role of the military. It appears worthwhile looking at the relationship between these tools both on the institutional as well as on the behavioral level. Finally, it is generally accepted that in many countries constitutional text widely diverges from constitutional reality. It is, therefore, unlikely that politicians always meticulously stick to the constraints laid down in the respective emergency provisions and it would be interesting to analyze differences between the de jure and the de facto emergency constitution more systematically. Such an analysis could begin by analysing to what degree the de jure provisions have been implemented in practice. For example, some constitutions mention that some “framework structure” is to be introduced. It is straightforward to ask whether it exists and if so, how many years elapsed between passing the constitution and creating the respective law. One could further compare the constitutionally mentioned maximum length of states of emergency with the length actually realized. Other potentially relevant aspects include: How many special courts have been established? What is the proportion of trials handled via such courts? Has the military been used domestically? Have actors involved in either the declaration or the implementation of a state of emergency been prosecuted after the dissolution of the state of emergency in case their behavior was not within the confines of legal action? Yet, any generally valid answer to all these questions must rely on some form of comparable operationalization of the features of emergency constitutions. As such, the findings in this article are arguably necessary to our ability to answer such questions in future research. We thank participants of the jubilee conference of the Eucken Institute (Freiburg, June 2014), the Public Choice Society Meeting (San Antonio, March 2015), the European Association of Law & Economics (Vienna, September 2015), and economic workshops at the Universities of Ankara (Bilkent), Bonn, Hamburg (Helmut Schmidt University), Lugano, Munich (TU), Prague (University of Economics), and Stellenbosch as well as Francisco Castellanos, Rola El Kabbani, Jerg Gutmann, Sven Höppner, Jean Michel Josselin, John Matsusaka, Stephan Michel, Konstantinos Pilpilidis, Agnes Strauss, Alois Stutzer, and Rok Spruk for valuable comments on earlier versions. All remaining errors are entirely ours. Footnotes 1 Zachary Elkins, Tom Ginsburg, & James Melton, The Endurance of National Constitutions (2009). 2 For the period between 1996 and 2004, Richards and Clay report thirty-five countries to have declared a state of emergency at least once, see David Richards & K. Chad Clay, An Umbrella with Holes: Respect for Non-Derogable Human Rights During Declared States of Emergency, 1996–2004, 13 Hum. Rts. Rev. 443 (2012). According to our search, we found a higher number, namely, forty-nine for the time period analyzed by Richards and Clay. Id. 3 Christian Davenport, Constitutional Promises and Repressive Reality: A Cross-National Time-Series Investigation of Why Political and Civil Liberties are Suppressed, 58(3) J. Pol. 627 (1996). 4 Linda Camp Keith & Steven C. Poe, Are Constitutional State of Emergency Clauses Effective? An Empirical Exploration, 26(4) Hum.Rts. Q. 1071 (2004). 5 Brian Loveman, The Constitution of Tyranny. Regimes of Exception in Spanish America (1993) is a very thorough analysis of the genesis of the emergency constitutions of many Latin American countries in the nineteenth century. 6 The text of the so-called constitution of the year three is documented at https://chnm.gmu.edu/revolution/d/450/. 7 All dates from Elkins et al., supra note 1. 8 It is, indeed, quite rare for emergency provisions to be added to a constitution by amendment. The only clear examples in democratic countries are Ireland in 1931 and Luxembourg in 2004; Germany provides a slightly different case as emergency provisions were added in 1968 to a constitution that had been devised while the country was still effectively occupied by Allied forces. A few other examples from non-democratic countries exist, in particular Comoros (2009), Cuba (1992), and the Soviet Union (1938). Bhutan is a special case in which emergency provisions were introduced in an interim constitution in 2005 that eventually carried over into the new constitution three years later. 9 Rafael La Porta, Florenzio Lopez-de-Silanes, & Andrei Shleifer, The Economic Consequences of Legal Origins, 46(2) J. Econ. Literature 285 (2008). 10 In addition to the countries already mentioned, these are El Salvador (1841), Haiti (1843), Mexico (1843), Dominican Republic (1844), Paraguay (1844), Guatemala (1845), Costa Rica (1848), and Honduras (1848). 11 Elkins et al., supra note 1. 12 The number of states also increased over this period. Expressed as a proportion, less than 60 percent of all states had an emergency constitution in 1850 whereas today, some 90 percent of all countries do. The source for our numbers is the Comparative Constitutions Project. The exact wording of the variable on which the numbers are based is: “Does the constitution have provisions for calling a state of emergency?” In each case where the question was answered in the affirmative, the country was coded as having an emergency constitution in the respective year. 13 According to the Comparative Constitutions Project, only seventeen countries have discarded constitutional emergency provisions. Elkins et al., supra note 1. 14 While an emergency constitution per se has not been re-introduced, the Austrian constitution has at least two clauses that could be called its emergency constitution: (i) art. 5(2) stipulates that in extraordinary times the president may issue decrees and (2) art. 79(2).2 that the military is to intervene as a consequence of natural disasters of extraordinary size. 15 The distinction between “new” and “amended” constitutions is not watertight. In our attempt to separate the two, we follow the classification of the Comparative Constitutions Project. Elkins et al., supra note 1. In the particular case of the Czech Republic, an amendment in 2009 allowed the parliament to dissolve itself under specific circumstances, which would include drastic emergencies. 16 Elkins et al., supra note 1. 17 These six potential justifications for declaring a state of emergency have been coded within the Comparative Constitutions Project. No further explanation on how to interpret them has been offered by those who created the dataset. We note that although some of these justifications are conceptually overlapping, we observe no emergency constitutions that, for example, mention both “general danger” and “threat to constitutional system” as separate items. 18 In principle, it is conceivable that states of emergencies are declared automatically after some pre-defined event occurred, such as an earthquake with a power of larger than 6 on the Richter scale. We are, however, not aware of any such automatism. 19 Throughout the article, we refer to this as separation of powers. Alternatively, one could also refer to the number of veto players, i.e., the number of actors whose consent is needed to make a specific decision. 20 An additional point that should ideally be included is who has the power to end a state of emergency, or who has the power to prolong an emergency. However, most emergency constitutions are surprisingly vague on this issue. We therefore refrain from including this aspect in the INEP, as the majority of the modern constitutions would have to be coded as having vague or no provisions. 21 Thorsten Beck et al., New Tools in Comparative Political Economy: The Database of Political Institutions, 15(1) World Bank Econ. Rev. 165 (2001). 22 Witold Jerzi Henisz, Political Constraint Index (POLCON) (2010). 23 Linda Camp Keith, Political Repression—Courts and the Law (2012). 24 According to the dataset from Linda Camp Keith which covers up to 181 countries. 25 Bruce Ackerman, The Emergency Constitution, 113(5) Yale L.J. 1029 (2004). 26 Keith,supra note 23. 27 John Ferejohn & Pasquale Pasquino, The Law of Exception: A Typology of Emergency Powers, 2 Int’l J. Const. L. 210, 216ff. (2004). 28 71 U.S. 2 (1866). 29 323 U.S. 214 (1944). 30 E.g., John Locke, The Work of John Locke in Nine Volumes. Volume 4: Economic Writings and Two Treatises of Government § 240 (1824). 31 Ackerman, supra note 25. 32 David Dyzenhaus, The Constitution of Law. Legality in a Time of Emergency (2006). 33 In the companion paper on consequences of emergency constitutions, we return to this issue by exploring the risk that emergencies lead to regime change, and in particular to loss of democratic institutions and violations of human rights. 34 Clinton L. Rossiter, Constitutional Dictatorship—Crisis Government in the Modern Democracies, with a new introduction by W.J. Quirk ch. IV (2009) gives an excellent account of that concept. It would be interesting to see whether the état de siège is followed more often by military regimes than other kinds of emergency concepts. Again, however, we delegate this question to a companion paper on consequences of emergency provisions. 35 This data is from the Comparative Constitutions Project again. 36 Keith,supra note 23. 37 In many cases, the CCP notes that central features of the emergency provisions are uncertain or unknown. We have rechecked many of these and only been able to fill out the blanks in the present German Constitution, the 1901 Danish Constitution, and the present constitution of Trinidad and Tobago. In all other cases we checked, we agreed with the CCP coders that not enough information was available to code the provisions. 38 The alternative would be to apply some form of multiplicative rule for how to aggregate the overall index. However, with a multiplicative index, one implicitly assumes that the single elements are complements such that, e.g., strict limits on declaration power reinforces the effects of having strict limits on the ability to suspend basic rights. We do not see any theoretical arguments for assuming that single elements are complements. Instead, our theoretical considerations lead us to believe that various provisions can substitute for each other, which necessitates applying and additive aggregation method. 39 Throughout, we code vague provisions as providing weaker emergency powers than explicit provisions providing some discretionary power. The power to derogate basic rights provides a pertinent example that we often observe in emergency constitutions. When the emergency constitution does allow some rights to be suspended, most provide an explicit list. When the constitution is vague on this issue, it does not contain a list. We therefore code vague provisions as a weaker provision than when a limited list is provided because the vague provisions may mean that some rights may be suspended—similarly to providing a limited list—and it may mean that they may not. 40 The dataset by Keith covers the period from 1979 until 2009 (Keith,supra note 23). During the first two thirds of this period, there was a slight increase in the separation of powers under states of emergency. However, this tendency is dominated by a change in the opposite direction over the entire period here under consideration, and particularly from 1950 until 1975 when decolonization was effectively through. 41 The overall trend describes an average of countries introducing substantially stronger emergency powers and countries weakening these powers. However, even when exploring the simple changes in the INEP, it is evident that about double the number of countries have increased rather than decreased the index while about half of the sample has had unchanged emergency constitutions since either 1950 or independence. 42 José Antonio Cheibub, Jennifer Gandhi, & James Raymond Vreeland, Democracy and Dictatorship Revisited, 143 (1–2) Pub. Choice 67 (2010). 43 E.g., Rossiter, supra note 34 (in particular ch. X). 44 Id. at 141. 45 Joseph F. Hair, Jr., et al., Multivariate Data Analysis (1998). 46 Countries without emergency constitutions are left out of this assignment. 47 A possible objection to analyzing both valid and defunct constitutions simultaneously could be that there might be some general trends that lead twenty-first-century emergency constitutions to be very different from nineteenth-century constitutions. If we reduce the analysis to only those constitutions currently in place, we lose around two-thirds of all our observations. Yet, clusters 1, 3, 5, and 6 remain stable whereas clusters 2 and 4 shift around a bit. 48 Which seems certainly in line with the fact that around one-third of them have been written by governments that came to power via a coup d’etat. 49 Supra note 42. 50 Supra note 22. 51 Angus Maddison, Statistics on World Population, GDP and Per Capita GDP, 1-2008 AD. Online database, available athttp://www.ggdc.net/MADDISON/oriindex.htm. 52 Our full dataset includes 351 constitutions, leaving 67 constitutions from countries with only one emergency constitution in the data. However, it must be noted that most of these countries have had more than one constitution although only one appears in the current data. The full CCP dataset, for example, includes information on four of Denmark’s five constitutions since 1849, but only the current 1953 constitution includes emergency provisions. 53 Elkins et al., supra note 1. 54 Christian Bjørnskov & Stefan Voigt, The Determinants of Emergency Constitutions (mimeo) (2016). 55 Christian Bjørnskov & Stefan Voigt, Why Do Governments Call a State of Emergency? On the Determinants of Using Emergency Constitutions, Eur. J. Pol. Econ. (forthcoming 2018). © The Author(s) 2018. Oxford University Press and New York University School of Law. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png International Journal of Constitutional Law Oxford University Press

The architecture of emergency constitutions

Loading next page...
 
/lp/ou_press/the-architecture-of-emergency-constitutions-TUNBNaFvkv
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Oxford University Press and New York University School of Law. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
1474-2640
eISSN
1474-2659
D.O.I.
10.1093/icon/moy012
Publisher site
See Article on Publisher Site

Abstract

Abstract Nine of ten countries currently have emergency provisions written into their constitutions, here simply referred to as emergency constitutions. The nature of these provisions remains poorly understood. We therefore aim at providing answers to two questions: (i) how much additional discretionary power do emergency constitutions allow and which political actors are given the additional power; and (ii) is there a limited number of “typical” emergency constitutions that combine various aspects in similar or even identical fashion? To answer the first question we construct an Indicator of Emergency Powers (INEP) which takes six central elements of emergency provisions explicitly into account. To answer the second question, we draw on cluster analysis and identify six well-defined clusters. Both the INEP as well as the six clusters allow us to answer important follow-up questions such as what the factors are that determine a country’s choice of emergency constitution but also under what conditions governments are likely to declare a state of emergency given the prevalent emergency constitution. 1. Introduction Today, some 90 percent of all constitutions worldwide contain explicit provisions for how to deal with states of emergency.1 The inclusion of emergency provisions—those legal rules specifying who can declare an emergency, when they can do so, and what actors have what powers once it has been declared—into constitutions has become the norm. They are used quite frequently and their use is often far from innocuous. Between 1985 and 2014, at least 137 countries declared a state of emergency at least once, thereby creating situations with particular political powers.2 Under a state of emergency, some individual rights and liberties are usually suspended and the separation of powers is curtailed in favor of the executive or even a single person like the head of state or government and, by implication, to the detriment of parliament and the courts. Assuming that states of emergency can be crucial moments for the development of entire countries and taking into consideration how frequently they are used, it is amazing how little we know about constitutional emergency provisions. Little is known about (i) the amount of additional powers granted to governments acting under a state of emergency, (ii) the trends in the evolution of emergency provisions over time, (iii) the factors that cause societies to adopt them in the first place, and (iv) their effects, i.e. the effectiveness in reaching the goals stated in the underlying legislation. Here, we focus exclusively on the first two questions, leaving the analysis of the latter questions to follow-up research. Cross-country studies analyzing the additional powers that emergency provisions grant their respective governments as well as the development of emergency provisions over time are extremely scarce. The existing cross-country studies rather focus on their consequences. Davenport,3 e.g., is interested in the determinants of state repression and asks whether the kinds of citizen rights mentioned in the constitution are significant for explaining variation in repressive behavior across governments. In that context, he asks whether explicitly listed limitations on citizen rights play a role and takes into account whether “martial law” and/or “state of emergency” are explicitly mentioned in the constitution. He further asks whether restrictions on either martial law or state of emergency are found. He is thus interested in neither the precise architecture of emergency constitutions nor their possible determinants. Keith and Poe4 ask whether emergency clauses are effective and are, hence, also concerned with their consequences and not their architecture.5 Due to this gap in the literature, we describe the main components of emergency constitutions as well as how they have changed over time. To make the powers that emergency constitutions allocate to governments comparable over time and countries, we develop an additive Index of Emergency Powers (INEP) that consists of six separate subindices. Quite often, legal scholars assert that emergency constitutions do not lend themselves to meaningful comparison. If this were true, questions dealing with the effectiveness of emergency constitutions would be almost impossible to answer. In this article, we ask whether this assumption is true and draw on cluster analysis to see whether we can find a limited number of clusters that share a number of traits. Drawing on 351 constitutions (both current and defunct) for which we have sufficient, available information and relying on thirty-one different variables capturing the most important features of emergency constitutions, we identify six such clusters and, hence, show that emergency constitutions often share many similarities. We find preliminary evidence that the consequences of emergency constitutions differ along the lines of these clusters. This result can now be used to ask a number of follow-up questions such as whether a particular type of emergency constitution is generally more effective than others, or whether we can identify specific situations in which a particular type is doing better than the other types and so on. The rest of the article is structured as follows: Section 2 names the main features of emergency constitutions and identifies time trends in their evolution. Section 3 briefly describes the main features of cluster analysis whereas Section 4 reports the results of the cluster analysis and a set of simple descriptive statistics. Section 5 concludes and spells out a number of follow-up questions that can be tackled with the INEP and the clusters here proposed. An online Appendix contains five tables which offer additional information such as precise definitions of the variables used, their sources and descriptive statistics. 2. Trends and emergency powers We define an emergency constitution as the set of formal legal provisions encoded in the constitution that specify who can declare an emergency, under which conditions an emergency can be declared, who needs to approve the declaration, and which actors have which special powers once it has been declared that the constitution does not assign to them outside emergencies. What we refer to as the “emergency constitution” here is, hence, not a document separate from the ordinary constitution but those formal provisions of it that explicitly deal with emergencies. We therefore do not capture decrees promulgated by government during a state of emergency considered to be part of the emergency constitution, conferral of emergency powers in statute, or any informal constitutional conventions or political tradition. Consequently, we consider the formal provisions literally written in the constitution as the outline of a legal maximum of government actions that are constitutionally allowed for. As such, emergency constitutions are paradoxical documents: Their declared goal is to re-establish constitutional order by temporarily suspending it. They are also paradoxical in the sense that the constitution spells out the conditions under which its regular application may be suspended. As such, emergency constitutions deal with the delicate balance between suspending individual rights by temporarily reducing the separation of powers, while also providing monitoring mechanisms intended to reduce the likelihood that the state of emergency is misused by power-maximizing politicians. Recent examples of countries calling a state of emergency include France, following the various terrorist attacks in Paris in November 2015, the May 2016 emergency declaration in Venezuela, and the October 2016 declaration in Ethiopia. These recent cases can serve to illustrate how emergency constitutions are applied and the large differences in which powers are available to the executive during emergencies. The French emergency constitution gives the authorities the right to derogate a number of basic rights, including the right to censor the press, yet the additional right mostly used has been the right of the police to search private homes without a warrant. Conversely, when Venezuelan president Nicolas Maduro declared a state of emergency in May 2016, he further restricted press freedom. When the government of Ethiopia in October 2016 declared a state of emergency following very large protests against the government, it also severely restricted information freedom. In both cases, the action was in violation of the countries’ own emergency constitutions, and the declaration in Venezuela ought to have been approved by the legislature before taking effect. In this section, we therefore start by giving a general overview of constitutional emergency provisions. We first look at their emergence and diffusion over time. We then deal with six central components of emergency provisions in detail. 2.1. On the diffusion of emergency constitutions Constitutional emergency provisions arguably do not make sense under absolute monarchy or totalitarian regimes. If the head of the executive is unconstrained, why should there be special provisions giving him powers he already enjoys? This is why the history of constitutional emergency provisions is closely linked with the advent of constitutional monarchy and with restricted government more generally. France was the first modern nation state to introduce constitutional emergency provisions in 1795.6 Emergency constitutions have since spread into those countries whose legal development was heavily influenced by France. In 1808, Joseph Bonaparte became king of Spain and the corresponding constitution—named after the French city of Bayonne because it was negotiated there—included explicit emergency provisions as in the French model. The provisions contained in the Bayonne constitution in turn served as a model for many Latin American constitutions. In fact, with the exception of Portugal (1826), all of the subsequent countries to create emergency constitutions in the following decades were Latin American. In chronological order, these were Argentina (1819), Chile (1822), Brazil (1824), United Provinces of the Rio de la Plata (1824), Bolivia (1826), Peru (1826), Ecuador (1830), Uruguay (1830), and Venezuela (1830).7 Two observations are noteworthy: first, in all of the cases mentioned, the inclusion of emergency provisions into the constitution was part of an entirely new constitution and, hence, not part of constitutional amendment.8 Second, all of these countries belong to the French legal family.9 By 1850, twenty countries had an emergency constitution; today, all of them are coded as “French legal origin.”10 But the spread of emergency constitutions did not stop in Latin America or countries of French legal origin. Using the extensive information contained in the Comparative Constitutions Project (CCP),11 we find that by 2013, 171 countries had in fact adopted emergency constitutions.12 This development is clearly illustrated in Figure 1, where we plot the share of independent countries with an emergency constitution, starting in 1900. The figure shows relative stability before 1950—the changes are mainly due to constitutions being abolished such that countries are coded as having no emergency constitution in the interim between abolishing the old constitution and either implementing a new constitution or re-instating the old. After 1950, two developments have clearly affected the share: the independence of former colonies, starting in the late 1950s, and the de facto independence of countries formerly part of or controlled by the Soviet Union. Figure 1. View largeDownload slide Share of constitutions with emergency provisions, 1900–2013. Figure 1. View largeDownload slide Share of constitutions with emergency provisions, 1900–2013. The data show that once a country has included emergency provisions into its constitutions, it is unlikely ever to get rid of them.13 Most countries that did get rid of their emergency constitution—often as a consequence of abolishing the constitution altogether—introduced a new one after a short interval. The only exception is Austria, which abolished its 1934 emergency constitution and has not re-introduced another to date.14 In all other cases, the abolishment was the consequence of a military or communist takeover that eventually resulted in the re-introduction of emergency provisions. As just described, the first constitutional emergency provisions in the nineteenth century were all part of entirely new constitutions. Virtually all of today’s emergency constitutions were already included in the last “new” constitution of a country. The 2002 amendment to the constituton of the Czech Republic is a notable exception in which emergency provisions were included later on.15 Hence, having an emergency constitution is now the rule rather than the exception. Yet, as we outline next, there is considerable variation with regards to specific details. 2.2. Single components of emergency provisions We suggest that every emergency constitution must, at least implicitly, deal with six different questions: a. What are the necessary conditions for a state of emergency? b. Who has the power to declare a state of emergency? c. Who has the power to declare the end of an emergency? d. Who has the power to monitor the legality of the means used during a state of emergency? e. Who exercises emergency powers? f. What (additional) competences does a state of emergency confer to the emergency government? (a) Necessary conditions Concerning the first dimension, the necessary preconditions for a state of emergency to be declared, two trends are noteworthy. First, emergency constitutions have become broader in the enumeration of events that can justify the declaration of an emergency. We rely on the variable in the Comparative Constitutions Project,16 which lists six possible preconditions for declaration: (i) war/aggression, (ii) internal security, (iii) national disaster, (iv) general danger, (v) economic emergency, and (vi) threat to constitutional system.17 Any of the given aspects might be found in a given proportion of constitutions, i.e., the share of constitutions including them can assume a value between 0 percent and 100 percent in any year. Adding up the resulting proportion for each of the six categories can, hence, yield a theoretical maximum of 600 (six categories of 100 percent). Albeit a crude measure, it does give an impression of how one aspect of emergency constitutions evolved over time: For 1950, the actual sum was 97.14 whereas the same exercise for 2011 yields a sum of 167.53. In other words, the possible causes for declaring a state of emergency have been considerably broadened and the typical, modern emergency constitution mentions two. Second, Table 1 shows that some preconditions have become more widely included than others. In 1950, 7.14 percent of all constitutions named any kind of “national disaster” as a potential justification for declaring a state of emergency, this proportion had grown to exactly one-third by 2011. “Economic emergencies,” on the other hand, were explicitly mentioned in 5.71 percent of all constitutions with emergency provisions in 1950. By 2013, this proportion had increased to only 7.73 percent. The sixth precondition in the list given above, “threat to the constitutional system,” did not exist in 1950, but by 2013 it was included in 4.12 percent of all constitutions with emergency provisions. Table 1. Percent of all emergency constitutions that name the respective topic as possible reason for calling a state of emergency. Topic 1950 2011 War/foreign aggresion 31.9 48.6 Internal security 31.9 38.8 National disaster 6.9 26.2 General danger 16.7 25.7 Economic emergency 5.6 7.1 Threat to constitutional system 0.0 2.7 Topic 1950 2011 War/foreign aggresion 31.9 48.6 Internal security 31.9 38.8 National disaster 6.9 26.2 General danger 16.7 25.7 Economic emergency 5.6 7.1 Threat to constitutional system 0.0 2.7 View Large Table 1. Percent of all emergency constitutions that name the respective topic as possible reason for calling a state of emergency. Topic 1950 2011 War/foreign aggresion 31.9 48.6 Internal security 31.9 38.8 National disaster 6.9 26.2 General danger 16.7 25.7 Economic emergency 5.6 7.1 Threat to constitutional system 0.0 2.7 Topic 1950 2011 War/foreign aggresion 31.9 48.6 Internal security 31.9 38.8 National disaster 6.9 26.2 General danger 16.7 25.7 Economic emergency 5.6 7.1 Threat to constitutional system 0.0 2.7 View Large (b) Power to declare The second component of emergency constitutions deals with the question of who has the power to declare an emergency. No matter how precisely the necessary preconditions are defined, some actors need to decide whether they are present or not.18 On the one hand, one can imagine a constitution that allocates the power to declare an emergency to the head of the executive branch without any other organ needing to approve of this decision. This would be equivalent to very few checks on executive power. On the other hand, an alternative scenario would be that more than one constitutional actor must be involved in the declaration and more than one actor must approve it.19 A third scenario, located somewhere between the two more extreme approaches just mentioned, could be that some other branch needs to be consulted; if its advice is not followed, this might increase the political costs of declaring an emergency in terms of reduced state legitimacy. Ideally, the necessary majorities should also be taken into account.20 Emergency constitutions typically reduce the degree of the separation of powers for a limited period of time. Yet, regarding non-emergency periods, there has been a secular trend toward a stricter separation of powers. The variable CHECKS, which proxies the degree of checks and balances and is part of the Database of Political Institutions,21 had a mean score of 1.2 in 1975 which had risen to 2.8 in 2012 (around a stable median of 2). Likewise, the mean score of Henisz’s22 PolConIII measure of political veto player power approximately doubled from 0.16 in 1950 to 0.3 in 2012, the latest year for which data is available. It seems possible then, that even emergency constitutions place more checks on the emergency government than they used to, if they follow the general constitutional trend. In the following we document that, in fact, the opposite has in general happened. Regarding the competence to declare a state of emergency, we observe two somewhat contradictory trends. In 1950, six of ten (22/37 = 59%) countries with an emergency constitution gave the head of state the right to declare a state of emergency. By 2011, this proportion had increased to 80 percent (129/159 = 81%) as most constitutions in the newly independent countries after the de-colonization wave of the 1950s and 1960s gave declaration rights to the head of state. This indicates the overwhelming importance of a single person, namely, the head of state. On the other hand, around 10 percent (16/159 = 10%) of all constitutions currently containing an emergency constitution allocate that competence to the entire cabinet. This is no increase in the separation of powers between legislature and executive, but at least an increase in the number of actors involved in the decision. Keith23 goes one step further and asks whether the responsibility for declaring a state of emergency is given explicitly to the legislature. In 1979, the first year of her dataset, this was the case in 4.2 percent of all countries. By 2010, this proportion had increased to 10 percent. If legislatures declare a state of emergency and executives are in charge during emergencies, then this change denotes an increase in the separation of powers. Another way to implement checks in emergency constitutions is to require the declaration of a state of emergency to be approved by an actor other than the one declaring it. In 1950, 44 percent of all emergency constitutions contained such a provision. By 2011, this proportion had increased to 56 percent, again indicating some increase in the level of checks. Today, consent of the following organs is most frequently required: (i) the first (or only) chamber of the legislature (39 percent); (ii) both chambers of the legislature (19 percent) and (iii) the government/cabinet (14 percent). Approval provisions are important because states of emergency have often been misused by self-serving politicians. In particular, some of them have simply dissolved the legislature to eliminate a watchdog. As of 2009, 15.5 percent of all emergency provisions explicitly excluded the possibility of dissolving the legislature during a state of emergency. In 1979, the respective number was only 3.9 percent.24 According to the CCP, no emergency constitution implemented before 1950 included an explicit ban on disolving the legislature. (c) Power to end The third important component of emergency constitutions is the power to end a state of emergency. One possibility—famously used by the Romans—is to have it expire automatically (after six months in Republican Rome). Here, too, the separation of powers is central. Acknowledging that every state of emergency entails the danger of misuse and the possibility of developing into a permanent autocracy, it seems reason able to allocate the power to end an emergency to an actor other than the one endowed with the exercise of emergency powers. Ackerman25 proposes a “supermajoritarian escalator”: the longer the state of emergency lasts, the more inclusive the parliamentary majority necessary to sustain it. Put differently, over time, ever smaller factions of parliament can end the state of emergency. Keith26 asks whether the emergency is constrained to a set time period and whether an extension is subject to legislative approval. In 1979, a little less than 19 percent of the 153 surveyed constitutions had such provisions, in 2009 the proportion had increased to 35.9 percent. At least de jure, many emergency constitutions thus try to sustain a high degree in their separation of powers by making extensions dependent on the consent of the legislature. (d) Power to monitor legality of actions under emergency Fourth, regarding actors who could monitor the legality of the means used under emergencies, both the legislature and the judiciary seem natural candidates. Ferejohn and Pasquino27 identify a “legislative model” of emergency powers which they distinguish from the (neo)-Roman model. In the legislative model, the legislature “is expected to monitor the use of emergency powers, to investigate abuses, to extend these powers if necessary, and perhaps to suspend them if the emergency ends.” They quote France as an example of a presidential system in which the legislature can impeach the president if it assesses that he has overstepped his powers (art. 16 of the French Constitution). As an example of legislative monitoring under a parliamentary system, they mention independent commissions in the UK installed to monitor the executive in the way it implements anti-terror legislation. Independent courts are a possible ex-post monitoring device. The US Supreme Court decisions Ex parte Milligan28 and Ex parte Korematsu29 are two (in)famous examples. Yet, many scholars traditionally believe that speed is of the essence in emergency situations and that judicial review should be postponed.30 Others, such as Ackerman31 or Dyzenhaus,32 point at the dismal record of the judiciary in constraining government action under emergency. Unfortunately, we are not aware of any dataset that would allow us to identify any time trend regarding this aspect.33 (e) Who exercises power? It seems natural to think of the head of the executive branch as the actor exercising emergency powers. Other provisions are, however, possible. They include the head of the military, but also technocrats. The French version of a state of emergency, the état de siège, implies an expansive delegation of powers to the military.34 Again, we are not aware of any dataset that would allow us to identify any change in this aspect over time. Finally, one must ask what competences are conferred onto the emergency government. First, emergency governments frequently entail the competence to suspend a number of basic rights. The proportion of countries whose constitutions provide for the suspension of rights during a state of emergency has remained virtually unchanged at 70 percent (30/43) in 1950 and 69 percent (118/172) in 2011.35 The dataset compiled by Keith36 analyzes the same issue from the opposite angle: according to her, the proportion of countries that do give a list of non-derogable rights or include a statement that certain rights cannot be infringed has changed from 5.2 percent in 1979 to almost 26 percent in 2010. Beyond the suspension of rights, there is a vast heterogeneity in the competences conferred on the emergency governments. Everything from “all powers necessary” to very detailed enumerations exists. Provisions frequently found include measures (i) to keep all government organs broadly conceived in office (vote of no confidence impossible, all elections suspended, terms of constitutional court judges extended), (ii) to increase the size of the army (including the use of national guards for military purposes), and (iii) to keep the country solvent (introducing new taxes, levying payment of existing taxes in advance) but also more dubious measures such as exempting state servants from all legal liability of state acts committed under martial law or enabling forced labor. Of course, there are also constitutions that expressly limit the competence of government under states of emergency, e.g., spelling out that no constitutional amendment can be passed under a state of emergency or that all decrees issued are only valid until the end of the state of emergency. 2.3. A new power index In order to synthesize these different aspects into a single dimension, we develop a measure that can be thought of as capturing the difficulty—or political cost—of calling and maintaining a state of emergency as well as its potential benefits. The simplest way to create an Index of Emergency Powers (INEP) is to rely on variables proxying for the most important aspects just described and add them up. Using the extensive information available from the Comparative Constitutions Project, this is exactly what we do. In the following, we therefore employ information covering up to 351 current as well as defunct constitutions for which we have been able to obtain sufficient information. Although the CCP naturally covers many more constitutions, central information of specific features in many now defunct constitutions is simply missing, which limits our sample to the present 351 examples.37 The INEP takes into account (i) the degree to which the right to declare a state of emergency is concentrated in a single person—or very few—or limited by multiple veto players; (ii) the need to and the degree to which this right is concentrated; (iii) how many different situations are explicitly mentioned in the constitution and can be used to justify the declaration of a state of emergency; (iv) whether fundamental civil and political rights can be suspended during a state of emergency; (v) whether parliament can be dissolved during a state of emergency; and (vi) whether the government can introduce censorship of the media and expropriate property during an emergency. The first three variables are, hence, concerned with the rules for declaring a state of emergency, whereas the last three are concerned with the powers that government enjoys under a state of emergency. The first three can also be thought of as “the cost element” of declaring a state of emergency whereas the last three cover “the benefits element” of running a state of emergency from the point of view of the incumbent government. In the INEP, higher coding in general implies more power to the executive. This is manifested in a lower degree of separation of powers captured in the first three components (the cost part of the INEP), where a high code implies a relative ease in declaring an emergency. For all elements of the index, no limits on the respective aspect is coded 3, some as 2, uncertainty as 1, and tight limits as 0. As we scale each of the six separate components on a 0–1 scale and subsequently scale the entire INEP on the same 0–1 scale, a coding of 1 would imply that there are no effective limits to the powers of the executive during emergencies and a coding of 0 that limits are maximally tight. The entire INEP is an additive index of the six separate components, as the different parts all represent different mechanisms allowed by the emergency constitution through which governments and leading political actors can directly affect decisions during emergencies. As such, a given level of checks and balances can, for example, be achieved by either limiting declaration power or approval power or perhaps including a sunset clause or a substantial limit on the powers given to the executive during an emergency. In other words, there is no clear progression across the six components, which would have necessitated a non-additive construction of an INEP. As such, due to the lack of a fine-grained theory that would inform us about the relative importance of each of those six components and their interplay, it seems straightforward to simply add the components up.38 In Table 2, we summarize the construction of the six components of the INEP. The components capture the power to declare a state of emergency and how concentrated it is, approval powers, conditions that are progressively more inclusive or vaguely defined, whether or not the legislature can be dissolved during emergencies, whether or not basic rights can be suspended during emergencies, and whether or not the constitution allows for expropriation of property and censorship.39 The INEP thus broadly consists of two main dimensions: (i) a cost dimension consisting of the first three components that outline the political costs and difficulty of declaring an emergency; and (ii) a benefit dimension that outlines the particular, discretionary powers that are allocated to the executive (or other government actors) during an emergency. These include the possibility to derogate basic rights, introduce censorship and expropriation without due compensation, and—for the executive—to dissolve parliament. Table 2. Constructing the INEP. Component Additive coding based on: Declaration power 2    if declaration rights rest with the head of government or the incumbent government; 1 if they are vaguely defined; 0 if they rest with the legislature or other (mainly courts) Approval power 3    if emergencies need no approval; 2 if approval rights rest with the head of government or the incumbent government; 1 if they are vaguely defined; 0 if they rest with the legislature or other (mainly courts) Conditions 3    if conditioned on “internal security” or “general danger”; 2 if they include “economic emergency” or “constitutional threat”; 1 if they include “other” or are vaguely defined; 0 if conditions are only “war” and “natural disasters” Dissolution power 1    if parliament can be dissolve during emergencies; 0 otherwise Rights suspension 3    if all rights can be suspended during emergencies; 2 if some can be suspended; 1 if the provisions are vague; 0 if no rights can be suspended Expropriation and censorship 1    if censorship can either be introduced during emergencies or is constitutionally allowed; 2 if authorities can expropriate without due compensation during emergencies Component Additive coding based on: Declaration power 2    if declaration rights rest with the head of government or the incumbent government; 1 if they are vaguely defined; 0 if they rest with the legislature or other (mainly courts) Approval power 3    if emergencies need no approval; 2 if approval rights rest with the head of government or the incumbent government; 1 if they are vaguely defined; 0 if they rest with the legislature or other (mainly courts) Conditions 3    if conditioned on “internal security” or “general danger”; 2 if they include “economic emergency” or “constitutional threat”; 1 if they include “other” or are vaguely defined; 0 if conditions are only “war” and “natural disasters” Dissolution power 1    if parliament can be dissolve during emergencies; 0 otherwise Rights suspension 3    if all rights can be suspended during emergencies; 2 if some can be suspended; 1 if the provisions are vague; 0 if no rights can be suspended Expropriation and censorship 1    if censorship can either be introduced during emergencies or is constitutionally allowed; 2 if authorities can expropriate without due compensation during emergencies View Large Table 2. Constructing the INEP. Component Additive coding based on: Declaration power 2    if declaration rights rest with the head of government or the incumbent government; 1 if they are vaguely defined; 0 if they rest with the legislature or other (mainly courts) Approval power 3    if emergencies need no approval; 2 if approval rights rest with the head of government or the incumbent government; 1 if they are vaguely defined; 0 if they rest with the legislature or other (mainly courts) Conditions 3    if conditioned on “internal security” or “general danger”; 2 if they include “economic emergency” or “constitutional threat”; 1 if they include “other” or are vaguely defined; 0 if conditions are only “war” and “natural disasters” Dissolution power 1    if parliament can be dissolve during emergencies; 0 otherwise Rights suspension 3    if all rights can be suspended during emergencies; 2 if some can be suspended; 1 if the provisions are vague; 0 if no rights can be suspended Expropriation and censorship 1    if censorship can either be introduced during emergencies or is constitutionally allowed; 2 if authorities can expropriate without due compensation during emergencies Component Additive coding based on: Declaration power 2    if declaration rights rest with the head of government or the incumbent government; 1 if they are vaguely defined; 0 if they rest with the legislature or other (mainly courts) Approval power 3    if emergencies need no approval; 2 if approval rights rest with the head of government or the incumbent government; 1 if they are vaguely defined; 0 if they rest with the legislature or other (mainly courts) Conditions 3    if conditioned on “internal security” or “general danger”; 2 if they include “economic emergency” or “constitutional threat”; 1 if they include “other” or are vaguely defined; 0 if conditions are only “war” and “natural disasters” Dissolution power 1    if parliament can be dissolve during emergencies; 0 otherwise Rights suspension 3    if all rights can be suspended during emergencies; 2 if some can be suspended; 1 if the provisions are vague; 0 if no rights can be suspended Expropriation and censorship 1    if censorship can either be introduced during emergencies or is constitutionally allowed; 2 if authorities can expropriate without due compensation during emergencies View Large Aggregating such issues into one index shows that emergency powers were quite limited in most constitutions for which we have information that were introduced before the Great Depression. A number of constitutions implemented in the 1930s, conversely, gave substantial powers to the executive, for instance, those of Yugoslavia (1931, INEP = 0.50), Poland (1935, INEP = 0.53), and Brazil (1937, INEP = 0.55). During the post-World War II period, as we illustrate in Figure 2, one can also observe clear developments. Several newly independent countries, for example, introduced emergency constitutions with strongly limited discretionary rights but rapidly increased those rights in amendments and new constitutions. Ghana’s 1957 constitution, with an INEP of 0.14, did not leave any declaration rights with the government and clearly delimited both approval rights and the conditions under which emergencies could be declared. Based on its 1969 constitution, Ghana’s INEP score increased to 0.51. The new constitution allocated all declaration powers to the head of government, allowed quite a few conditions under which to declare an emergency, and allowed basic rights to be suspended during emergencies. The evolution of Ghana’s constitution is, overall, representative of the general development of former colonies as they became independent and implemented constitutions of their own. Figure 2. View largeDownload slide Average INEP, 1950–2010. Figure 2. View largeDownload slide Average INEP, 1950–2010. While there was a slight tendency for emergency constitutions to be more delimiting during the late 1980s and 1990s,40 in more recent years a number of new constitutions allowed for substantial unchecked emergency powers to the executive. New constitutions in Kenya (2010, INEP = 0.55), Guinea (2010, INEP = 0.66), and Hungary (2011, INEP = 0.71) all leave significant discretionary power to the executive during broadly defined states of emergency. These developments can be clearly seen as an uptick in Figure 2.41 Although they seem to coincide with the time after the 9/11 attacks, it is unclear whether the attacks caused the uptick in the figure as it is driven by constitutional changes in countries not directly affected by terrorist threats. Separating the six components of the INEP, as we do in Figure 3, reveals that the main development in the early years after World War II is a concentration of declaration rights. Similarly, the figure shows that the new constitutions, implemented in former colonies between 1960 and the late 1970s, also were more likely to allow for the dissolution of the legislature. Conversely, constitutions implemented after the collapse of communism clearly tend to be less likely to allow expropriation and censorship but likely to include more conditions under which a state of emergency can be declared. Figure 3. View largeDownload slide Six separate indices, INEP, 1950–2010 Figure 3. View largeDownload slide Six separate indices, INEP, 1950–2010 Overall, as we show in Table 3, there is no substantial difference between democracies and non-democracies regarding emergency powers—using the DD (democracy and dictatorship) index indicator developed by Cheibub et al.,42 the average INEP value is 0.31 in both groups. The DD data also allow for the separation of three types of democracy (parliamentary, presidential, and mixed) and three types of dictatorship (civil, military, and royal). In the data, only royal dictatorships in our sample have much stronger declaration rights—other autocracies do not—while presidential democracies tend to include more conditions under which a state of emergency can be declared. Historically mixed democracies, i.e., democracies with mostly ceremonial presidents without actual powers, have been less prone to allow suspension of rights, expropriation and censorship, but in the present emergency constitutions, all types of democracies now appear similar on average. Table 3. Characteristics of emergency constitutions, regime types. Democracy Autocracy Parliamentary Mixed Presidential Civil Military Royal Overall INEP 0.32 0.26 0.33 0.29 0.33 0.30 Declaration power 0.66 0.71 0.69 0.71 0.78 0.85 Approval power 0.21 0.18 0.19 0.21 0.23 0.17 Conditions 0.29 0.24 0.44 0.23 0.30 0.21 Dissolution power 0.18 0.13 0.08 0.09 0.09 0.03 Rights suspension 0.19 0.08 0.25 0.15 0.18 0.08 Expropriation and censorship 0.39 0.25 0.32 0.36 0.43 0.46 Democracy Autocracy Parliamentary Mixed Presidential Civil Military Royal Overall INEP 0.32 0.26 0.33 0.29 0.33 0.30 Declaration power 0.66 0.71 0.69 0.71 0.78 0.85 Approval power 0.21 0.18 0.19 0.21 0.23 0.17 Conditions 0.29 0.24 0.44 0.23 0.30 0.21 Dissolution power 0.18 0.13 0.08 0.09 0.09 0.03 Rights suspension 0.19 0.08 0.25 0.15 0.18 0.08 Expropriation and censorship 0.39 0.25 0.32 0.36 0.43 0.46 View Large Table 3. Characteristics of emergency constitutions, regime types. Democracy Autocracy Parliamentary Mixed Presidential Civil Military Royal Overall INEP 0.32 0.26 0.33 0.29 0.33 0.30 Declaration power 0.66 0.71 0.69 0.71 0.78 0.85 Approval power 0.21 0.18 0.19 0.21 0.23 0.17 Conditions 0.29 0.24 0.44 0.23 0.30 0.21 Dissolution power 0.18 0.13 0.08 0.09 0.09 0.03 Rights suspension 0.19 0.08 0.25 0.15 0.18 0.08 Expropriation and censorship 0.39 0.25 0.32 0.36 0.43 0.46 Democracy Autocracy Parliamentary Mixed Presidential Civil Military Royal Overall INEP 0.32 0.26 0.33 0.29 0.33 0.30 Declaration power 0.66 0.71 0.69 0.71 0.78 0.85 Approval power 0.21 0.18 0.19 0.21 0.23 0.17 Conditions 0.29 0.24 0.44 0.23 0.30 0.21 Dissolution power 0.18 0.13 0.08 0.09 0.09 0.03 Rights suspension 0.19 0.08 0.25 0.15 0.18 0.08 Expropriation and censorship 0.39 0.25 0.32 0.36 0.43 0.46 View Large Tracing the development of emergency constitutions over time, we thus find both general patterns and substantial differences across countries. Emergency constitutions have become more prevalent while the discretionary powers that they confer on the executive continue to vary. 3. Cluster analysis In the last section, we mentioned that emergency constitutions first evolved in France and then spread across the countries belonging to the French legal family. Some scholars propose that the type of emergency constitution brought about within the French legal family is different from the one brought about within the common law legal family. The French approach is referred to as the state of siege (état de siège) whereas the British one would originate from martial law.43 Two aspects in particular make them distinct from each other, namely, (ii) the identity of the actor authorized to monitor the state of emergency and (ii) the degree of judicialization. Under a state of siege, monitoring would primarily fall to the legislature, whereas it would fall on the judiciary, and more precisely the regular courts, under martial law. An important consequence of this difference is that monitoring by the legislature can take place during an emergency while monitoring by the courts will only take place after the state of emergency has been ended. Other than for the state of siege, martial law could be characterized by “the absence of statutory foresight for its initiation and use.”44 This last trait makes the identification of martial law via the analysis of formal constitutions somewhat difficult. We now ask whether this dichotomy regarding types of emergency constitutions is reflected in the data, i.e., whether there are clear “families” of emergency constitutions. In order to assess whether one can identify families of emergency constitutions, we employ cluster analysis. To generate clusters, we rely on exactly the same variables used for creating the INEP. Cluster analysis is a set of methods that, at its most basic, aims at minimizing the distance in n-dimensional space between member observations within a cluster while maximizing the distance between clusters. In other words, cluster analysis is a way to identify families of observations that are as similar as possible while being distinctly different form other families. We use the K-means technique with random centroids from which a K-means algorithm optimizes the Euclidean distance within and between clusters.45 As we do not have strong priors about the correct or merely reasonable number of clusters, we perform a set of analyses defining between two and ten clusters. Based on this set of analyses, we select what we believe is the number of clusters that best fits the data. Our criteria are that (i) intra-cluster distances ought to be relatively small—i.e. we require clusters to be fairly coherent; (ii) distances between clusters ought to be comparatively large; and (iii) no single cluster has disproportionately high intra-cluster distances or large outlier members. This last criterion ensures that we do not obtain clusters that are “residual” clusters only containing observations that do not fit into any other cluster. We employ these criteria using graphs similar to standard screen plots in which the gain in identification can be easily eyeballed. In the following, we outline the details of these analyses before describing our preferred solution. 4. Are there typical emergency constitutions? 4.1. The number of constitutional clusters In Table 4, we outline some pertinent features of each of the first nine K-means cluster solutions we obtain. Table 4. Cluster solution specifics. No. of clusters Average intra-cluster distance 2 3 4 5 6 7 8 9 Cluster 1 3.05 2.92 3.31 3.05 2.81 2.22 2.09 2.58 Cluster 2 3.33 3.11 2.88 2.88 2.37 2.53 2.74 1.89 Cluster 3 3.18 2.69 2.24 2.31 3.27 2.29 2.69 Cluster 4 2.92 2.92 2.11 2.39 2.62 2.65 Cluster 5 2.97 3.19 3.44 2.22 2.95 Cluster 6 3.49 2.25 3.46 4.91 Cluster 7 3.09 2.44 2.11 Cluster 8 3.05 2.43 Cluster 9 3.56 “Goodness of fit” measures Av. distance 3.19 3.07 2.96 2.81 2.72 2.74 2.61 2.83 Min. distance 3.05 2.92 2.69 2.24 2.11 2.22 2.09 1.89 Max. distance 3.33 3.18 3.31 3.06 3.49 3.44 3.46 4.91 Largest outlier 8.012 8.34 7.72 7.35 7.13 7.25 7.54 7.36 Montenegro (2007) Germany (1949) Hungary (2011) Germany (1949) Germany (1949) Germany (1949) Hungary (2011) Montenegro (2007) No. of clusters Average intra-cluster distance 2 3 4 5 6 7 8 9 Cluster 1 3.05 2.92 3.31 3.05 2.81 2.22 2.09 2.58 Cluster 2 3.33 3.11 2.88 2.88 2.37 2.53 2.74 1.89 Cluster 3 3.18 2.69 2.24 2.31 3.27 2.29 2.69 Cluster 4 2.92 2.92 2.11 2.39 2.62 2.65 Cluster 5 2.97 3.19 3.44 2.22 2.95 Cluster 6 3.49 2.25 3.46 4.91 Cluster 7 3.09 2.44 2.11 Cluster 8 3.05 2.43 Cluster 9 3.56 “Goodness of fit” measures Av. distance 3.19 3.07 2.96 2.81 2.72 2.74 2.61 2.83 Min. distance 3.05 2.92 2.69 2.24 2.11 2.22 2.09 1.89 Max. distance 3.33 3.18 3.31 3.06 3.49 3.44 3.46 4.91 Largest outlier 8.012 8.34 7.72 7.35 7.13 7.25 7.54 7.36 Montenegro (2007) Germany (1949) Hungary (2011) Germany (1949) Germany (1949) Germany (1949) Hungary (2011) Montenegro (2007) Note: average, minimum and maximum distances reported in the lower panel are inter-cluster distances. The largest outlier is the distance of the given country to the centroid of the cluster in which it is placed. View Large Table 4. Cluster solution specifics. No. of clusters Average intra-cluster distance 2 3 4 5 6 7 8 9 Cluster 1 3.05 2.92 3.31 3.05 2.81 2.22 2.09 2.58 Cluster 2 3.33 3.11 2.88 2.88 2.37 2.53 2.74 1.89 Cluster 3 3.18 2.69 2.24 2.31 3.27 2.29 2.69 Cluster 4 2.92 2.92 2.11 2.39 2.62 2.65 Cluster 5 2.97 3.19 3.44 2.22 2.95 Cluster 6 3.49 2.25 3.46 4.91 Cluster 7 3.09 2.44 2.11 Cluster 8 3.05 2.43 Cluster 9 3.56 “Goodness of fit” measures Av. distance 3.19 3.07 2.96 2.81 2.72 2.74 2.61 2.83 Min. distance 3.05 2.92 2.69 2.24 2.11 2.22 2.09 1.89 Max. distance 3.33 3.18 3.31 3.06 3.49 3.44 3.46 4.91 Largest outlier 8.012 8.34 7.72 7.35 7.13 7.25 7.54 7.36 Montenegro (2007) Germany (1949) Hungary (2011) Germany (1949) Germany (1949) Germany (1949) Hungary (2011) Montenegro (2007) No. of clusters Average intra-cluster distance 2 3 4 5 6 7 8 9 Cluster 1 3.05 2.92 3.31 3.05 2.81 2.22 2.09 2.58 Cluster 2 3.33 3.11 2.88 2.88 2.37 2.53 2.74 1.89 Cluster 3 3.18 2.69 2.24 2.31 3.27 2.29 2.69 Cluster 4 2.92 2.92 2.11 2.39 2.62 2.65 Cluster 5 2.97 3.19 3.44 2.22 2.95 Cluster 6 3.49 2.25 3.46 4.91 Cluster 7 3.09 2.44 2.11 Cluster 8 3.05 2.43 Cluster 9 3.56 “Goodness of fit” measures Av. distance 3.19 3.07 2.96 2.81 2.72 2.74 2.61 2.83 Min. distance 3.05 2.92 2.69 2.24 2.11 2.22 2.09 1.89 Max. distance 3.33 3.18 3.31 3.06 3.49 3.44 3.46 4.91 Largest outlier 8.012 8.34 7.72 7.35 7.13 7.25 7.54 7.36 Montenegro (2007) Germany (1949) Hungary (2011) Germany (1949) Germany (1949) Germany (1949) Hungary (2011) Montenegro (2007) Note: average, minimum and maximum distances reported in the lower panel are inter-cluster distances. The largest outlier is the distance of the given country to the centroid of the cluster in which it is placed. View Large Following the results in the table toward the right, it is easy to see that the intra-cluster distance across the solutions decreases until the six-cluster solution, to the right of which it increases again. The minimum distance—indicating the coherence of a cluster—also decreases until the six-cluster solution. Likewise, the largest outlier in the six-cluster solution, i.e., the constitution placed the furthest away from other cluster members in 31-dimensional space, is relatively the smallest. Moving beyond a solution with six clusters entails a deterioration in at least one of our “goodness of fit” measures and in particular leaves larger outliers and in all cases at least one comparatively poorly defined cluster, as captured by the average distance. In all cases, we find that three particular constitutions fit any family of constitutions poorly: emergency provisions of the present constitutions of Germany (the 1949 constitution as amended in 1968), Hungary, and Montenegro are structurally different from most other emergency constitutions. The overall intra-cluster coherence on the basis of six clusters is also fairly good in comparison to alternative cluster numbers. This is a final reason for choosing six clusters. Based on the proposed criteria, a solution with six clusters best fit the present data.46Table 5 reports the full cluster membership of this solution for all 351 constitutions in the dataset.47 Table 5. Cluster memberships, preferred six-cluster solution. Cluster Members 1. Domestic security Afghanistan (2004), Albania (1918, 1998), Angola (2010), Antigua and Barbuda (1981), Argentina (1853), Azerbaijan (1995), Barbados (1966), Belarus (1994), Bolivia (1938, 1947, 1961, 1967, 2009), Brazil (1967), Cambodia (1971), Chile (1833, 1980), Congo (1991, 2001), DR Congo (1964), Costa Rica (1917), Cyprus (1960), Ecuador (1869, 1897, 1946, 1967, 1978, 1984, 1998), Eritrea (1997), Estonia (1991), Ethiopia (1994), Ghana (1979, 1991), Guatemala (1945, 1965), Haiti (1889, 1950, 1964, 1983, 1987), Honduras (1965, 1981), Iran (1979), Kenya (2010), Kosovo (2008), Kyrgyzstan (2010), Liberia (1986), Lithuania (1921), Macedonia (1991), Madagascar (1991), Malawi (1994), Maldives (2008), Mexico (1857, 1917), Mongolia (1991), Mozambique (2004), Myanmar (2008), Namibia (1990), Nepal (1990), Nicaragua (1987), Niger (1960), Nigeria (1989, 1999), Panama (1971), Paraguay (1991), Peru (1979), Philippines 1973), Poland (1921, 1935, 1997), Portugal (1976), Samoa (1961), Saudi Arabia (1991), Seychelles (1993), Spain (1931), St. Kitts and Nevis (1983), St. Vincent and the Grenadines (1979), Swaziland (2005), Syria (1953), Tajikistan (1994), Thailand (1968, 1997, 2007), Turkey (1924, 1945, 1961, 1981), Uganda (1995), Uruguay (1918), Uzbekistan (1991), Vanuatu (1980), Venezuela (1999) 2. Middle of the road Afghanistan (1964), Albania (1915), Benin (1964), Bolivia (1851), Bulgaria (1991), Burundi (1981, 1991, 2005), Cameroon (1961), Central African Republic (1981, 2004), China (1913, 1954), Colombia (1886), Comoros (1978), Congo (1963), Cote D’Ivoire (2000), East Timor (2001), Egypt (1971), Estonia (1937), Ethiopia (1955, 1987), Fiji (1970, 1990, 1997), Finland (1999), France (1851), Gabon (1961, 1975), Gambia (1970), Greece (1951), Guatemala (1879), Guinea (1981), Haiti (1843, 1946), Kazakhstan (1995), Kenya (1963), Kiribati (1979), Kuwait (1961), Kyrgyzstan (1993), Latvia (1921), Lithuania (1918, 1938), Madagascar (1998), Maldives (1998), Mali (1974), Mauritania (1991), Mongolia (1960), Morocco (1961), Mozambique (1990), Myanmar (1974), Nauru (1968), Nepal (1948, 1959, 1961), Niger (1996, 1999), Pakistan (1961), Peru (1818), Philippines (1987), Poland (1951, 1991), Qatar (2003), Senegal (2001), Sierra Leone (1978), Solomon Islands (1978), Spain (1978), Togo (1963, 1991), Tuvalu (1986), Venezuela (1961), South Vietnam (1956), Zimbabwe (1979) 3. Serious on rights Albania (1976), Andorra (1993), Bolivia (1816), Bulgaria (1947, 1971), Burkina Faso (1960), Burundi (1974), Central African Republic (1994), Chad (1961), Denmark (1953), Dominican Republic (1947), Equatorial Guinea (1981), Germany (1871), Ghana (1957), Guinea (1990), Honduras (1936), Japan (1889), Jordan (1946), Laos (1991), Madagascar (1975), Mali (1991), Mauritania (1961), Moldova (1994), Mozambique (1975), Nicaragua (1905, 1948, 1950, 1974), Niger (1991), Peru (1867), Somalia (1979), South Africa (1961, 1983), Syria (1950, 1973), Thailand (1949), Togo (1979), Trinidad and Tobago (1961), USA (1789), Venezuela (1904, 1947,1953), Vietnam (1980, 1991), Yugoslavia (1953) 4. Current design Afghanistan (1977, 1990), Albania (1924), Bahamas (1973), Bahrain (2001), Benin (1990), Botswana (1966), Cambodia (1989), Chile (1811), China (1981), Congo (1979), DR Congo (2003), Costa Rica (1949), Djibouti (1991), Egypt (1923, 2011), Equatorial Guinea (1991), Estonia (1920), Fiji (2013), Gambia (1996), Guinea (2010), Guyana (1970), Indonesia (1945), Iraq (2005), Jordan (1951), Kyrgyzstan (2007), Lesotho (1966), Libya (1951), Malawi (1964), Poland (1947), Russia (1993), Rwanda (1961), Senegal (1960, 1963), Seychelles (1979), Soviet Union (1977), Sri Lanka (1978), Sudan (1973, 1998), Surinam (1987), Thailand (1974), Uganda (1967), Ukraine (1996), North Vietnam (1960), Arab Republic of Yemen (1970), Zambia (1964, 1973, 1991), Zimbabwe (1969, 2013) 5. Strong legislature Argentina (1819, 1826), Austria (1934), Brazil (1891, 1946), Chile (1925), Costa Rica (1869), Cuba (1935, 1940), Dominican Republic (1896, 1908, 1914, 1917, 1955, 1961a, 1961b, 1963, 1966, 1994, 2001), El Salvador (1886, 1939, 1950, 1983), France (1946), West Germany (1949), Haiti (1807), Honduras (1880, 1904, 1914, 1957), Hungary (2011), Lithuania (1991), Montenegro (2007), Nicaragua (1893, 1911), Nigeria (1960, 1963), Panama (1904, 1946), Paraguay (1870), Peru (1860, 1910), Portugal (1933), Serbia and Montenegro (2003, 2006), Sierra Leone (1961), South Africa (1993, 1996), Venezuela (1874, 1881, 1891), Yugoslavia (1921) 6. Rights suspenders Armenia (1995), Belize (1981); Bhutan (2008), Bolivia (1878, 1880), Brazil (1937), Burkina Faso (1991), Cameroon (1971), Central African Republic (1964, 1976), Chad (1960, 1996), Comoros (1996), DR Congo (1978, 2005), Cuba (1959), Ecuador (1884, 1906, 1993, 1996, 2008), France (1848, 1958), Ghana (1969), Jamaica (1961), Lesotho (1993), Mauritania (1978), Morocco (1970, 2011), Niger (1989), Oman (1996), Paraguay (1940), Rwanda (2003), Sri Lanka (1971), Yugoslavia (1931, 1974) Cluster Members 1. Domestic security Afghanistan (2004), Albania (1918, 1998), Angola (2010), Antigua and Barbuda (1981), Argentina (1853), Azerbaijan (1995), Barbados (1966), Belarus (1994), Bolivia (1938, 1947, 1961, 1967, 2009), Brazil (1967), Cambodia (1971), Chile (1833, 1980), Congo (1991, 2001), DR Congo (1964), Costa Rica (1917), Cyprus (1960), Ecuador (1869, 1897, 1946, 1967, 1978, 1984, 1998), Eritrea (1997), Estonia (1991), Ethiopia (1994), Ghana (1979, 1991), Guatemala (1945, 1965), Haiti (1889, 1950, 1964, 1983, 1987), Honduras (1965, 1981), Iran (1979), Kenya (2010), Kosovo (2008), Kyrgyzstan (2010), Liberia (1986), Lithuania (1921), Macedonia (1991), Madagascar (1991), Malawi (1994), Maldives (2008), Mexico (1857, 1917), Mongolia (1991), Mozambique (2004), Myanmar (2008), Namibia (1990), Nepal (1990), Nicaragua (1987), Niger (1960), Nigeria (1989, 1999), Panama (1971), Paraguay (1991), Peru (1979), Philippines 1973), Poland (1921, 1935, 1997), Portugal (1976), Samoa (1961), Saudi Arabia (1991), Seychelles (1993), Spain (1931), St. Kitts and Nevis (1983), St. Vincent and the Grenadines (1979), Swaziland (2005), Syria (1953), Tajikistan (1994), Thailand (1968, 1997, 2007), Turkey (1924, 1945, 1961, 1981), Uganda (1995), Uruguay (1918), Uzbekistan (1991), Vanuatu (1980), Venezuela (1999) 2. Middle of the road Afghanistan (1964), Albania (1915), Benin (1964), Bolivia (1851), Bulgaria (1991), Burundi (1981, 1991, 2005), Cameroon (1961), Central African Republic (1981, 2004), China (1913, 1954), Colombia (1886), Comoros (1978), Congo (1963), Cote D’Ivoire (2000), East Timor (2001), Egypt (1971), Estonia (1937), Ethiopia (1955, 1987), Fiji (1970, 1990, 1997), Finland (1999), France (1851), Gabon (1961, 1975), Gambia (1970), Greece (1951), Guatemala (1879), Guinea (1981), Haiti (1843, 1946), Kazakhstan (1995), Kenya (1963), Kiribati (1979), Kuwait (1961), Kyrgyzstan (1993), Latvia (1921), Lithuania (1918, 1938), Madagascar (1998), Maldives (1998), Mali (1974), Mauritania (1991), Mongolia (1960), Morocco (1961), Mozambique (1990), Myanmar (1974), Nauru (1968), Nepal (1948, 1959, 1961), Niger (1996, 1999), Pakistan (1961), Peru (1818), Philippines (1987), Poland (1951, 1991), Qatar (2003), Senegal (2001), Sierra Leone (1978), Solomon Islands (1978), Spain (1978), Togo (1963, 1991), Tuvalu (1986), Venezuela (1961), South Vietnam (1956), Zimbabwe (1979) 3. Serious on rights Albania (1976), Andorra (1993), Bolivia (1816), Bulgaria (1947, 1971), Burkina Faso (1960), Burundi (1974), Central African Republic (1994), Chad (1961), Denmark (1953), Dominican Republic (1947), Equatorial Guinea (1981), Germany (1871), Ghana (1957), Guinea (1990), Honduras (1936), Japan (1889), Jordan (1946), Laos (1991), Madagascar (1975), Mali (1991), Mauritania (1961), Moldova (1994), Mozambique (1975), Nicaragua (1905, 1948, 1950, 1974), Niger (1991), Peru (1867), Somalia (1979), South Africa (1961, 1983), Syria (1950, 1973), Thailand (1949), Togo (1979), Trinidad and Tobago (1961), USA (1789), Venezuela (1904, 1947,1953), Vietnam (1980, 1991), Yugoslavia (1953) 4. Current design Afghanistan (1977, 1990), Albania (1924), Bahamas (1973), Bahrain (2001), Benin (1990), Botswana (1966), Cambodia (1989), Chile (1811), China (1981), Congo (1979), DR Congo (2003), Costa Rica (1949), Djibouti (1991), Egypt (1923, 2011), Equatorial Guinea (1991), Estonia (1920), Fiji (2013), Gambia (1996), Guinea (2010), Guyana (1970), Indonesia (1945), Iraq (2005), Jordan (1951), Kyrgyzstan (2007), Lesotho (1966), Libya (1951), Malawi (1964), Poland (1947), Russia (1993), Rwanda (1961), Senegal (1960, 1963), Seychelles (1979), Soviet Union (1977), Sri Lanka (1978), Sudan (1973, 1998), Surinam (1987), Thailand (1974), Uganda (1967), Ukraine (1996), North Vietnam (1960), Arab Republic of Yemen (1970), Zambia (1964, 1973, 1991), Zimbabwe (1969, 2013) 5. Strong legislature Argentina (1819, 1826), Austria (1934), Brazil (1891, 1946), Chile (1925), Costa Rica (1869), Cuba (1935, 1940), Dominican Republic (1896, 1908, 1914, 1917, 1955, 1961a, 1961b, 1963, 1966, 1994, 2001), El Salvador (1886, 1939, 1950, 1983), France (1946), West Germany (1949), Haiti (1807), Honduras (1880, 1904, 1914, 1957), Hungary (2011), Lithuania (1991), Montenegro (2007), Nicaragua (1893, 1911), Nigeria (1960, 1963), Panama (1904, 1946), Paraguay (1870), Peru (1860, 1910), Portugal (1933), Serbia and Montenegro (2003, 2006), Sierra Leone (1961), South Africa (1993, 1996), Venezuela (1874, 1881, 1891), Yugoslavia (1921) 6. Rights suspenders Armenia (1995), Belize (1981); Bhutan (2008), Bolivia (1878, 1880), Brazil (1937), Burkina Faso (1991), Cameroon (1971), Central African Republic (1964, 1976), Chad (1960, 1996), Comoros (1996), DR Congo (1978, 2005), Cuba (1959), Ecuador (1884, 1906, 1993, 1996, 2008), France (1848, 1958), Ghana (1969), Jamaica (1961), Lesotho (1993), Mauritania (1978), Morocco (1970, 2011), Niger (1989), Oman (1996), Paraguay (1940), Rwanda (2003), Sri Lanka (1971), Yugoslavia (1931, 1974) View Large Table 5. Cluster memberships, preferred six-cluster solution. Cluster Members 1. Domestic security Afghanistan (2004), Albania (1918, 1998), Angola (2010), Antigua and Barbuda (1981), Argentina (1853), Azerbaijan (1995), Barbados (1966), Belarus (1994), Bolivia (1938, 1947, 1961, 1967, 2009), Brazil (1967), Cambodia (1971), Chile (1833, 1980), Congo (1991, 2001), DR Congo (1964), Costa Rica (1917), Cyprus (1960), Ecuador (1869, 1897, 1946, 1967, 1978, 1984, 1998), Eritrea (1997), Estonia (1991), Ethiopia (1994), Ghana (1979, 1991), Guatemala (1945, 1965), Haiti (1889, 1950, 1964, 1983, 1987), Honduras (1965, 1981), Iran (1979), Kenya (2010), Kosovo (2008), Kyrgyzstan (2010), Liberia (1986), Lithuania (1921), Macedonia (1991), Madagascar (1991), Malawi (1994), Maldives (2008), Mexico (1857, 1917), Mongolia (1991), Mozambique (2004), Myanmar (2008), Namibia (1990), Nepal (1990), Nicaragua (1987), Niger (1960), Nigeria (1989, 1999), Panama (1971), Paraguay (1991), Peru (1979), Philippines 1973), Poland (1921, 1935, 1997), Portugal (1976), Samoa (1961), Saudi Arabia (1991), Seychelles (1993), Spain (1931), St. Kitts and Nevis (1983), St. Vincent and the Grenadines (1979), Swaziland (2005), Syria (1953), Tajikistan (1994), Thailand (1968, 1997, 2007), Turkey (1924, 1945, 1961, 1981), Uganda (1995), Uruguay (1918), Uzbekistan (1991), Vanuatu (1980), Venezuela (1999) 2. Middle of the road Afghanistan (1964), Albania (1915), Benin (1964), Bolivia (1851), Bulgaria (1991), Burundi (1981, 1991, 2005), Cameroon (1961), Central African Republic (1981, 2004), China (1913, 1954), Colombia (1886), Comoros (1978), Congo (1963), Cote D’Ivoire (2000), East Timor (2001), Egypt (1971), Estonia (1937), Ethiopia (1955, 1987), Fiji (1970, 1990, 1997), Finland (1999), France (1851), Gabon (1961, 1975), Gambia (1970), Greece (1951), Guatemala (1879), Guinea (1981), Haiti (1843, 1946), Kazakhstan (1995), Kenya (1963), Kiribati (1979), Kuwait (1961), Kyrgyzstan (1993), Latvia (1921), Lithuania (1918, 1938), Madagascar (1998), Maldives (1998), Mali (1974), Mauritania (1991), Mongolia (1960), Morocco (1961), Mozambique (1990), Myanmar (1974), Nauru (1968), Nepal (1948, 1959, 1961), Niger (1996, 1999), Pakistan (1961), Peru (1818), Philippines (1987), Poland (1951, 1991), Qatar (2003), Senegal (2001), Sierra Leone (1978), Solomon Islands (1978), Spain (1978), Togo (1963, 1991), Tuvalu (1986), Venezuela (1961), South Vietnam (1956), Zimbabwe (1979) 3. Serious on rights Albania (1976), Andorra (1993), Bolivia (1816), Bulgaria (1947, 1971), Burkina Faso (1960), Burundi (1974), Central African Republic (1994), Chad (1961), Denmark (1953), Dominican Republic (1947), Equatorial Guinea (1981), Germany (1871), Ghana (1957), Guinea (1990), Honduras (1936), Japan (1889), Jordan (1946), Laos (1991), Madagascar (1975), Mali (1991), Mauritania (1961), Moldova (1994), Mozambique (1975), Nicaragua (1905, 1948, 1950, 1974), Niger (1991), Peru (1867), Somalia (1979), South Africa (1961, 1983), Syria (1950, 1973), Thailand (1949), Togo (1979), Trinidad and Tobago (1961), USA (1789), Venezuela (1904, 1947,1953), Vietnam (1980, 1991), Yugoslavia (1953) 4. Current design Afghanistan (1977, 1990), Albania (1924), Bahamas (1973), Bahrain (2001), Benin (1990), Botswana (1966), Cambodia (1989), Chile (1811), China (1981), Congo (1979), DR Congo (2003), Costa Rica (1949), Djibouti (1991), Egypt (1923, 2011), Equatorial Guinea (1991), Estonia (1920), Fiji (2013), Gambia (1996), Guinea (2010), Guyana (1970), Indonesia (1945), Iraq (2005), Jordan (1951), Kyrgyzstan (2007), Lesotho (1966), Libya (1951), Malawi (1964), Poland (1947), Russia (1993), Rwanda (1961), Senegal (1960, 1963), Seychelles (1979), Soviet Union (1977), Sri Lanka (1978), Sudan (1973, 1998), Surinam (1987), Thailand (1974), Uganda (1967), Ukraine (1996), North Vietnam (1960), Arab Republic of Yemen (1970), Zambia (1964, 1973, 1991), Zimbabwe (1969, 2013) 5. Strong legislature Argentina (1819, 1826), Austria (1934), Brazil (1891, 1946), Chile (1925), Costa Rica (1869), Cuba (1935, 1940), Dominican Republic (1896, 1908, 1914, 1917, 1955, 1961a, 1961b, 1963, 1966, 1994, 2001), El Salvador (1886, 1939, 1950, 1983), France (1946), West Germany (1949), Haiti (1807), Honduras (1880, 1904, 1914, 1957), Hungary (2011), Lithuania (1991), Montenegro (2007), Nicaragua (1893, 1911), Nigeria (1960, 1963), Panama (1904, 1946), Paraguay (1870), Peru (1860, 1910), Portugal (1933), Serbia and Montenegro (2003, 2006), Sierra Leone (1961), South Africa (1993, 1996), Venezuela (1874, 1881, 1891), Yugoslavia (1921) 6. Rights suspenders Armenia (1995), Belize (1981); Bhutan (2008), Bolivia (1878, 1880), Brazil (1937), Burkina Faso (1991), Cameroon (1971), Central African Republic (1964, 1976), Chad (1960, 1996), Comoros (1996), DR Congo (1978, 2005), Cuba (1959), Ecuador (1884, 1906, 1993, 1996, 2008), France (1848, 1958), Ghana (1969), Jamaica (1961), Lesotho (1993), Mauritania (1978), Morocco (1970, 2011), Niger (1989), Oman (1996), Paraguay (1940), Rwanda (2003), Sri Lanka (1971), Yugoslavia (1931, 1974) Cluster Members 1. Domestic security Afghanistan (2004), Albania (1918, 1998), Angola (2010), Antigua and Barbuda (1981), Argentina (1853), Azerbaijan (1995), Barbados (1966), Belarus (1994), Bolivia (1938, 1947, 1961, 1967, 2009), Brazil (1967), Cambodia (1971), Chile (1833, 1980), Congo (1991, 2001), DR Congo (1964), Costa Rica (1917), Cyprus (1960), Ecuador (1869, 1897, 1946, 1967, 1978, 1984, 1998), Eritrea (1997), Estonia (1991), Ethiopia (1994), Ghana (1979, 1991), Guatemala (1945, 1965), Haiti (1889, 1950, 1964, 1983, 1987), Honduras (1965, 1981), Iran (1979), Kenya (2010), Kosovo (2008), Kyrgyzstan (2010), Liberia (1986), Lithuania (1921), Macedonia (1991), Madagascar (1991), Malawi (1994), Maldives (2008), Mexico (1857, 1917), Mongolia (1991), Mozambique (2004), Myanmar (2008), Namibia (1990), Nepal (1990), Nicaragua (1987), Niger (1960), Nigeria (1989, 1999), Panama (1971), Paraguay (1991), Peru (1979), Philippines 1973), Poland (1921, 1935, 1997), Portugal (1976), Samoa (1961), Saudi Arabia (1991), Seychelles (1993), Spain (1931), St. Kitts and Nevis (1983), St. Vincent and the Grenadines (1979), Swaziland (2005), Syria (1953), Tajikistan (1994), Thailand (1968, 1997, 2007), Turkey (1924, 1945, 1961, 1981), Uganda (1995), Uruguay (1918), Uzbekistan (1991), Vanuatu (1980), Venezuela (1999) 2. Middle of the road Afghanistan (1964), Albania (1915), Benin (1964), Bolivia (1851), Bulgaria (1991), Burundi (1981, 1991, 2005), Cameroon (1961), Central African Republic (1981, 2004), China (1913, 1954), Colombia (1886), Comoros (1978), Congo (1963), Cote D’Ivoire (2000), East Timor (2001), Egypt (1971), Estonia (1937), Ethiopia (1955, 1987), Fiji (1970, 1990, 1997), Finland (1999), France (1851), Gabon (1961, 1975), Gambia (1970), Greece (1951), Guatemala (1879), Guinea (1981), Haiti (1843, 1946), Kazakhstan (1995), Kenya (1963), Kiribati (1979), Kuwait (1961), Kyrgyzstan (1993), Latvia (1921), Lithuania (1918, 1938), Madagascar (1998), Maldives (1998), Mali (1974), Mauritania (1991), Mongolia (1960), Morocco (1961), Mozambique (1990), Myanmar (1974), Nauru (1968), Nepal (1948, 1959, 1961), Niger (1996, 1999), Pakistan (1961), Peru (1818), Philippines (1987), Poland (1951, 1991), Qatar (2003), Senegal (2001), Sierra Leone (1978), Solomon Islands (1978), Spain (1978), Togo (1963, 1991), Tuvalu (1986), Venezuela (1961), South Vietnam (1956), Zimbabwe (1979) 3. Serious on rights Albania (1976), Andorra (1993), Bolivia (1816), Bulgaria (1947, 1971), Burkina Faso (1960), Burundi (1974), Central African Republic (1994), Chad (1961), Denmark (1953), Dominican Republic (1947), Equatorial Guinea (1981), Germany (1871), Ghana (1957), Guinea (1990), Honduras (1936), Japan (1889), Jordan (1946), Laos (1991), Madagascar (1975), Mali (1991), Mauritania (1961), Moldova (1994), Mozambique (1975), Nicaragua (1905, 1948, 1950, 1974), Niger (1991), Peru (1867), Somalia (1979), South Africa (1961, 1983), Syria (1950, 1973), Thailand (1949), Togo (1979), Trinidad and Tobago (1961), USA (1789), Venezuela (1904, 1947,1953), Vietnam (1980, 1991), Yugoslavia (1953) 4. Current design Afghanistan (1977, 1990), Albania (1924), Bahamas (1973), Bahrain (2001), Benin (1990), Botswana (1966), Cambodia (1989), Chile (1811), China (1981), Congo (1979), DR Congo (2003), Costa Rica (1949), Djibouti (1991), Egypt (1923, 2011), Equatorial Guinea (1991), Estonia (1920), Fiji (2013), Gambia (1996), Guinea (2010), Guyana (1970), Indonesia (1945), Iraq (2005), Jordan (1951), Kyrgyzstan (2007), Lesotho (1966), Libya (1951), Malawi (1964), Poland (1947), Russia (1993), Rwanda (1961), Senegal (1960, 1963), Seychelles (1979), Soviet Union (1977), Sri Lanka (1978), Sudan (1973, 1998), Surinam (1987), Thailand (1974), Uganda (1967), Ukraine (1996), North Vietnam (1960), Arab Republic of Yemen (1970), Zambia (1964, 1973, 1991), Zimbabwe (1969, 2013) 5. Strong legislature Argentina (1819, 1826), Austria (1934), Brazil (1891, 1946), Chile (1925), Costa Rica (1869), Cuba (1935, 1940), Dominican Republic (1896, 1908, 1914, 1917, 1955, 1961a, 1961b, 1963, 1966, 1994, 2001), El Salvador (1886, 1939, 1950, 1983), France (1946), West Germany (1949), Haiti (1807), Honduras (1880, 1904, 1914, 1957), Hungary (2011), Lithuania (1991), Montenegro (2007), Nicaragua (1893, 1911), Nigeria (1960, 1963), Panama (1904, 1946), Paraguay (1870), Peru (1860, 1910), Portugal (1933), Serbia and Montenegro (2003, 2006), Sierra Leone (1961), South Africa (1993, 1996), Venezuela (1874, 1881, 1891), Yugoslavia (1921) 6. Rights suspenders Armenia (1995), Belize (1981); Bhutan (2008), Bolivia (1878, 1880), Brazil (1937), Burkina Faso (1991), Cameroon (1971), Central African Republic (1964, 1976), Chad (1960, 1996), Comoros (1996), DR Congo (1978, 2005), Cuba (1959), Ecuador (1884, 1906, 1993, 1996, 2008), France (1848, 1958), Ghana (1969), Jamaica (1961), Lesotho (1993), Mauritania (1978), Morocco (1970, 2011), Niger (1989), Oman (1996), Paraguay (1940), Rwanda (2003), Sri Lanka (1971), Yugoslavia (1931, 1974) View Large 4.2. The cluster solution Although cluster analysis can at times yield solutions that are difficult to interpret, the present six-cluster solution in the table indicates that the clusters are all defined by traits common to their specific member constitutions. Cluster 1 thus appears to be defined by a concern for domestic security,48 cluster 3 by having particularly well-defined rights protection, while cluster 4 contains many relatively recent constitutions and thus represents what may be termed current design. Cluster 2, on the other hand, is characterized by emergency constitutions that on most counts represent an average constitutional design. Finally, the average within-country distances indicate that clusters 5 and 6 may be less clearly defined than the four first clusters (based on information contained in Table 4). We nevertheless find that constitutions allocated to clusters 5 and 6 share particular identifiable features. Overall, it turns out that most of the differences in the various types of emergency constitutions can be traced back to four variables, namely, (i) legislative declaration, i.e., the role of the legislature in declaring a state of emergency; (ii) the specific approval powers; (iii) the specific conditions that allow declaring a state of emergency; and (4) which rights, if any, can be suspended during emergencies. We therefore interpret these final clusters as defined by having strong legislatures and easy rights suspension, respectively. We proceed with a number of noteworthy observations summarized in Table 6. Table 6. Cluster characteristics. 1 2 3 4 5 6 Domesticsecurity Middle of the road Serious on rights Current design Strong legislature Easy rights suspenders Centroid Guatemala 1945 Congo 1963 Bulgaria 1947 Zimbabwe 1969 Paraguay 1870 Ecuador 1906 Av. distance 2.81 2.37 2.31 2.11 3.19 3.49 Members 94 73 45 50 53 36 INEP 0.36 0.33 0.26 0.34 0.23 0.49 Declaration power 0.94 0.92 0.64 0.95 0.07 Legislative declaration 0.94 Approval power 0.13 Legislative approval 0.29 0.24 Vague approval prov. 0.21 0.38 0.29 Other approval Conditions 0.50 Internal security 0.23 Nat. disasters not mentioned 0.15 War, nat. disasters not mentioned 0.13 Vague conditions 0.39 0.39 Dissolution power 0.07 0.08 0.04 0.20 0.06 0.31 Rights suspension 0.21 Rights suspension 0.07 0.06 0.12 .23 Rights suspension 0.56 Some rights suspension Expropriation and censorship 0.32 0.40 Censorship 0.40 Censorship 0.44 0.27 0.44 Censorship Age 45 51 62 42 84 48 Length 18,706 13,138 8,727 15,800 13,466 15,226 Legal origin 83% civil 79% civil 67% civil 60% civil 91% civil 81% civil Democracy 37% 25% 20% 14% 38% 33% PolConIII 0.10 0.098 0.063 0.065 0.12 0.19 GDP at birth 2,905 2,565 2,062 2,443 2,361 2,268 Still in use 18 6 6 22 8 13 Post-Cold War 39% 29% 18% 34% 17% 36% 1 2 3 4 5 6 Domesticsecurity Middle of the road Serious on rights Current design Strong legislature Easy rights suspenders Centroid Guatemala 1945 Congo 1963 Bulgaria 1947 Zimbabwe 1969 Paraguay 1870 Ecuador 1906 Av. distance 2.81 2.37 2.31 2.11 3.19 3.49 Members 94 73 45 50 53 36 INEP 0.36 0.33 0.26 0.34 0.23 0.49 Declaration power 0.94 0.92 0.64 0.95 0.07 Legislative declaration 0.94 Approval power 0.13 Legislative approval 0.29 0.24 Vague approval prov. 0.21 0.38 0.29 Other approval Conditions 0.50 Internal security 0.23 Nat. disasters not mentioned 0.15 War, nat. disasters not mentioned 0.13 Vague conditions 0.39 0.39 Dissolution power 0.07 0.08 0.04 0.20 0.06 0.31 Rights suspension 0.21 Rights suspension 0.07 0.06 0.12 .23 Rights suspension 0.56 Some rights suspension Expropriation and censorship 0.32 0.40 Censorship 0.40 Censorship 0.44 0.27 0.44 Censorship Age 45 51 62 42 84 48 Length 18,706 13,138 8,727 15,800 13,466 15,226 Legal origin 83% civil 79% civil 67% civil 60% civil 91% civil 81% civil Democracy 37% 25% 20% 14% 38% 33% PolConIII 0.10 0.098 0.063 0.065 0.12 0.19 GDP at birth 2,905 2,565 2,062 2,443 2,361 2,268 Still in use 18 6 6 22 8 13 Post-Cold War 39% 29% 18% 34% 17% 36% View Large Table 6. Cluster characteristics. 1 2 3 4 5 6 Domesticsecurity Middle of the road Serious on rights Current design Strong legislature Easy rights suspenders Centroid Guatemala 1945 Congo 1963 Bulgaria 1947 Zimbabwe 1969 Paraguay 1870 Ecuador 1906 Av. distance 2.81 2.37 2.31 2.11 3.19 3.49 Members 94 73 45 50 53 36 INEP 0.36 0.33 0.26 0.34 0.23 0.49 Declaration power 0.94 0.92 0.64 0.95 0.07 Legislative declaration 0.94 Approval power 0.13 Legislative approval 0.29 0.24 Vague approval prov. 0.21 0.38 0.29 Other approval Conditions 0.50 Internal security 0.23 Nat. disasters not mentioned 0.15 War, nat. disasters not mentioned 0.13 Vague conditions 0.39 0.39 Dissolution power 0.07 0.08 0.04 0.20 0.06 0.31 Rights suspension 0.21 Rights suspension 0.07 0.06 0.12 .23 Rights suspension 0.56 Some rights suspension Expropriation and censorship 0.32 0.40 Censorship 0.40 Censorship 0.44 0.27 0.44 Censorship Age 45 51 62 42 84 48 Length 18,706 13,138 8,727 15,800 13,466 15,226 Legal origin 83% civil 79% civil 67% civil 60% civil 91% civil 81% civil Democracy 37% 25% 20% 14% 38% 33% PolConIII 0.10 0.098 0.063 0.065 0.12 0.19 GDP at birth 2,905 2,565 2,062 2,443 2,361 2,268 Still in use 18 6 6 22 8 13 Post-Cold War 39% 29% 18% 34% 17% 36% 1 2 3 4 5 6 Domesticsecurity Middle of the road Serious on rights Current design Strong legislature Easy rights suspenders Centroid Guatemala 1945 Congo 1963 Bulgaria 1947 Zimbabwe 1969 Paraguay 1870 Ecuador 1906 Av. distance 2.81 2.37 2.31 2.11 3.19 3.49 Members 94 73 45 50 53 36 INEP 0.36 0.33 0.26 0.34 0.23 0.49 Declaration power 0.94 0.92 0.64 0.95 0.07 Legislative declaration 0.94 Approval power 0.13 Legislative approval 0.29 0.24 Vague approval prov. 0.21 0.38 0.29 Other approval Conditions 0.50 Internal security 0.23 Nat. disasters not mentioned 0.15 War, nat. disasters not mentioned 0.13 Vague conditions 0.39 0.39 Dissolution power 0.07 0.08 0.04 0.20 0.06 0.31 Rights suspension 0.21 Rights suspension 0.07 0.06 0.12 .23 Rights suspension 0.56 Some rights suspension Expropriation and censorship 0.32 0.40 Censorship 0.40 Censorship 0.44 0.27 0.44 Censorship Age 45 51 62 42 84 48 Length 18,706 13,138 8,727 15,800 13,466 15,226 Legal origin 83% civil 79% civil 67% civil 60% civil 91% civil 81% civil Democracy 37% 25% 20% 14% 38% 33% PolConIII 0.10 0.098 0.063 0.065 0.12 0.19 GDP at birth 2,905 2,565 2,062 2,443 2,361 2,268 Still in use 18 6 6 22 8 13 Post-Cold War 39% 29% 18% 34% 17% 36% View Large Let us begin with those constitutions not grouped in any of the six clusters because they do not have emergency constitutions. We identify fifty-eight constitutions from twenty-one countries as such constitutions. On average, these constitutions are older, shorter, and less likely to be located in a civil law system and all those constitutions containing emergency provisions. Of the twelve constitutions without emergency provisions that remain in use today, five belong to systems coded as democratic according to Cheibub et al.49 These democracies are in no way clearly different from democracies in which the constitutions include emergency provisions. The second noteworthy observation is that the mean INEP for clusters 1, 2 and 4 is fairly similar at around 0.35 whereas it is substantially lower in clusters 3 and 5 (around 0.25) and much larger in clusters 6 at 0.49. However, as is evident in Table 6, the regular constitutions in countries in cluster 6 also have the strongest veto player institutions, as measured by Henisz’s50 (2010) PolConIII index. As such, regardless of the mix of characteristics, cluster 6 seems to be characterized by a substantially larger difference between the separation of powers in ordinary versus emergency times than other clusters. We also note that constitutions in cluster 1 are, in general, substantially longer and thus more detailed than in other clusters whereas constitutions in cluster 3 are shorter. Conversely, comparing legal origins and real PPP (purchasing power parity)-adjusted GDP (gross domestic product) per capita at the time when the constitution was introduced, which we take from the Maddison database,51 we find no discernible differences in legal origins or GDP, while constitutions in cluster 3 tend to be older and very few constitutions in cluster 4 were written when the country was democratic. This is a rather unexpected finding given the history of emergency powers described in Section 2 of this article. We capture the particular mix of powers in an alternative way in the multinomial logistic regression reported in Table 7, which clearly shows that—relative to the baseline of cluster 1—emergency provisions in clusters 3 and 5 are substantially and significantly more restricted than in the remaining clusters. Clusters 1 and 4 stand out with weaker approval powers than the rest although it is also clear that cluster 1 is characterized by having legislative approval while approval in cluster 6 is left to “other,” meaning non-political actors. Regarding the right to suspend basic rights, this is strongly circumscribed in cluster 3 and broadly allowed in constitutions in clusters 2 and 6. Finally, clusters 3, 4, and 6 are characterized by allowing censorship and expropriation without ordinary compensation during emergencies. Table 7. Determinants of cluster placement, constitutional characteristics Cluster 2 3 4 5 6 Declaration power −0.506 (0.796) −2.682*** (0.789) 0.402 (0.992) −7.835*** (1.159) −0.156 (0.979) Approval power 2.329*** (0.745) 2.156** (0.912) 1.329 (0.880) 5.632*** (1.326) 2.515*** (0.853) Conditions −6.874*** (1.169) −10.903*** (1.716) −12.848*** (1.774) 0.528 (1.682) −2.999*** (1.131) Dissolution power −0.427 (0.655) −1.668* (0.993) 0.508 (0.666) −1.366 (1.019) 0.929 (0.643) Rights suspension 2.958*** (1.058) −4.279*** (1.545) −1.099 (1.045) −2.547 (1.617) 3.267*** (0.745) Expropriation and censorship 1.633* (0.863) 2.292** (1.115) 2.697** (1.118) 2.234* (1.263) 1.964* (1.085) Observations 351 LR chi squared 490.76 Pseudo R squared 0.402 Cluster 2 3 4 5 6 Declaration power −0.506 (0.796) −2.682*** (0.789) 0.402 (0.992) −7.835*** (1.159) −0.156 (0.979) Approval power 2.329*** (0.745) 2.156** (0.912) 1.329 (0.880) 5.632*** (1.326) 2.515*** (0.853) Conditions −6.874*** (1.169) −10.903*** (1.716) −12.848*** (1.774) 0.528 (1.682) −2.999*** (1.131) Dissolution power −0.427 (0.655) −1.668* (0.993) 0.508 (0.666) −1.366 (1.019) 0.929 (0.643) Rights suspension 2.958*** (1.058) −4.279*** (1.545) −1.099 (1.045) −2.547 (1.617) 3.267*** (0.745) Expropriation and censorship 1.633* (0.863) 2.292** (1.115) 2.697** (1.118) 2.234* (1.263) 1.964* (1.085) Observations 351 LR chi squared 490.76 Pseudo R squared 0.402 *Denotes significance at p < .10. ** Denotes significance at p < .05. *** Denotes significance at p < .01. View Large Table 7. Determinants of cluster placement, constitutional characteristics Cluster 2 3 4 5 6 Declaration power −0.506 (0.796) −2.682*** (0.789) 0.402 (0.992) −7.835*** (1.159) −0.156 (0.979) Approval power 2.329*** (0.745) 2.156** (0.912) 1.329 (0.880) 5.632*** (1.326) 2.515*** (0.853) Conditions −6.874*** (1.169) −10.903*** (1.716) −12.848*** (1.774) 0.528 (1.682) −2.999*** (1.131) Dissolution power −0.427 (0.655) −1.668* (0.993) 0.508 (0.666) −1.366 (1.019) 0.929 (0.643) Rights suspension 2.958*** (1.058) −4.279*** (1.545) −1.099 (1.045) −2.547 (1.617) 3.267*** (0.745) Expropriation and censorship 1.633* (0.863) 2.292** (1.115) 2.697** (1.118) 2.234* (1.263) 1.964* (1.085) Observations 351 LR chi squared 490.76 Pseudo R squared 0.402 Cluster 2 3 4 5 6 Declaration power −0.506 (0.796) −2.682*** (0.789) 0.402 (0.992) −7.835*** (1.159) −0.156 (0.979) Approval power 2.329*** (0.745) 2.156** (0.912) 1.329 (0.880) 5.632*** (1.326) 2.515*** (0.853) Conditions −6.874*** (1.169) −10.903*** (1.716) −12.848*** (1.774) 0.528 (1.682) −2.999*** (1.131) Dissolution power −0.427 (0.655) −1.668* (0.993) 0.508 (0.666) −1.366 (1.019) 0.929 (0.643) Rights suspension 2.958*** (1.058) −4.279*** (1.545) −1.099 (1.045) −2.547 (1.617) 3.267*** (0.745) Expropriation and censorship 1.633* (0.863) 2.292** (1.115) 2.697** (1.118) 2.234* (1.263) 1.964* (1.085) Observations 351 LR chi squared 490.76 Pseudo R squared 0.402 *Denotes significance at p < .10. ** Denotes significance at p < .05. *** Denotes significance at p < .01. View Large As such, a simple and preliminary cluster analysis reveals that one can to some extent identify “ideal” types of emergency constitution design. As such, although we do not want to overstate the degree of familiarity within each cluster, the standard claim that each emergency constitution is a unique document that does not lend itself to easy comparison seems at least questionable. To sum up, we observe that emergency constitutions can be separated into six clusters. These clusters differ not only in the make-up of the emergency provisions and the specific degree powers given to the executive, but also in their likelihood of declaring a state of emergency if an actual emergency arises. It is with this background that we now turn to our theoretical considerations and an intuitive test of whether the distribution of countries within the cluster families is stable over time. 4.3. How stable is the cluster assignment for countries over time? As a final element in describing the families of emergency constitutions, we provide information on how stable a country’s placement in a particular cluster or constitutional family is. This exercise requires that we can observe constitutional changes in the emergency provisions within countries. The data include 284 constitutions from eighty-one countries on which we have sufficient information.52 The data first of all shows that eight countries represented by more than one emergency constitution in the data have not changed clusters and therefore constitutional families: El Salvador (four emergency constitutions), Fiji (three), Gabon (two), Mexico (two), Russia and the Soviet Union (two), Sudan (two), Turkey (four), and Zambia (three). Their constitutional choices of emergency provisions remain quite constant over the years due to either political tradition, geographical circumstance, or mere coincidence. Conversely, of the fifty-three countries with at least three constitutions in the dataset, twenty-three had emergency constitutions represented in three different families and Albania, Bolivia, Niger, Peru, and Venezuela have all had emergency constitutions belonging to four different families. However, many of these constitutional changes of course occur between families that are not particularly different. We therefore further note that comparing the Euclidean distance between cluster centroids, the largest relative changes occur between clusters 1 and 3 and clusters 4 and 5. Conversely, clusters 1 and 2 and clusters 2 and 6 are more similar than other cluster pairs. One should therefore not overestimate the importance of emergency constitutions changing constitutional families. The countries with the relatively largest changes are: Albania in clusters 1 (1918 and 1998) and 3 (1976), Bolivia in clusters 3 (1816) and 1 (all twentieth-century constitutions); Chile in clusters 4 (1811) and 5 (1925); Costa Rica in clusters 4 (1949) and 5 (1869); Ghana in clusters 3 (1957) and 1 (subsequent constitutions); Honduras in clusters 3 (1936) and 1 (1965 and 1981); Madagascar in clusters 3 (1975) and 1 (1991); and Mozambique in clusters 1 (2004) and 3 (1975). Nicaragua, whose 1987 constitution in cluster 1 was also substantially different from its other emergency constitutions (all in cluster 3); Niger in clusters 1 (1960) and 3 (1991); Peru in clusters 3 (1867) and 1 (1979); Thailand whose 1949 constitution in cluster 3 was substantially different from its subsequent emergency constitutions; and Venezuela in which Hugo Chavez’s 1999 socialist constitution in cluster 1 had very different emergency provisions than previous constitutions (all in cluster 3). We do not want to delve deeper into the reasons for these changes except to note two regularities. First, in the cases of the 1991 Bulgarian and the 2003 Serbian emergency constitutions, it is evident that they both represented a constitutional return to the constitutional family that their constitutions prior to communism had belonged to. The second regularity is that Latin American countries are substantially more likely to have experienced larger changes to their emergency constitutions. While these countries have also had the most constitutional changes, 47 of the 100 emergency constitutions in the twenty Latin American countries with recorded changes have occurred in countries that have changed constitutional family. Eleven of the twenty countries have even had constitutions in at least three clusters. In sum, identifying a limited number of “typical” emergency constitutions seems very well possible via cluster analysis. Generally, the content of many constitutions changes only marginally even after regime changes.53 This is different with regard to emergency constitutions. Here, shifts between clusters are not at all uncommon. 5. Conclusions and outlook Given that nine out of ten constitutions contain emergency provisions, it is amazing how little we know about them. This article tries to reduce our ignorance by doing two things. First, it contains an Index of Emergency Powers, which measures the degree of discretionary power constitutionally allocated to the executive during emergencies. On average, the overall degree in the separation of powers outlined by constitutions has increased over the last decades. Relying on the INEP, we find that this is emphatically not true for emergency constitutions. Quite to the contrary, we observe a clear long-run trend to allocate more, rather than less, powers to the executive during times of emergency. Our data suggests that this trend has been a consequence of the relatively weak separation of powers in the constitutions of newly independent countries since the late 1950s, and the particular features of constitutions in the countries that became de facto independent after the collapse of the communist bloc in the early 1990s. Second, this article deals with the presumption that emergency constitutions are unique and do not lend themselves to straightforward comparisons with other emergency constitutions. Drawing on thirty-one different variables and employing cluster analysis, we are able to identify six clusters in which we group 351 different constitutions for which we have sufficient information. These clusters each define what can be thought of as a “family” of emergency constitutions that are sufficiently similar to be characterized by very comparable constitutional choices. The main features that separate these families of emergency constitutions are choices regarding the role of the legislature in declaring a state of emergency, the specific approval powers, the constitutional conditions that allow declaring a state of emergency, and which basic individual rights, if any, can be suspended during emergencies. This article is interested in making emergency constitutions comparable. The indicator of emergency powers (the INEP) as well as the clusters here introduced promise to be relevant for a number of highly topical follow-up questions. First, what are the factors that determine what kind of emergency constitution a country adopts, given that it adopts one? Is being prone to natural disasters as prominent a factor as one might expect or are other, less salient ones, even more important? Bjørnskov and Voigt54 is a first attempt to answer these questions. Second, one might ask what the determinants of actually declaring a state of emergency are. Can one show that emergency constitutions entailing more separation of powers are used less frequently? What about the use of states of emergency in countries without explicit emergency provisions? Bjørnskov and Voigt55 deal with these questions. Third, what are the effects of emergency constitutions? Given that we have identified six different components of emergency powers, can one identify one as particularly apt in re-establishing the status quo ante? Or, in minimizing the number of dead after a natural disaster has occurred? Or as inadequate in re-establishing ex ante levels of civil rights? How do countries without an explicit emergency constitution fare in comparison? Fourth, emergency constitutions are but one tool that can serve the interests of power-maximizing politicians. Other tools include, but are not restricted to, the frequent use of executive decrees, the creation of special courts, and the role of the military. It appears worthwhile looking at the relationship between these tools both on the institutional as well as on the behavioral level. Finally, it is generally accepted that in many countries constitutional text widely diverges from constitutional reality. It is, therefore, unlikely that politicians always meticulously stick to the constraints laid down in the respective emergency provisions and it would be interesting to analyze differences between the de jure and the de facto emergency constitution more systematically. Such an analysis could begin by analysing to what degree the de jure provisions have been implemented in practice. For example, some constitutions mention that some “framework structure” is to be introduced. It is straightforward to ask whether it exists and if so, how many years elapsed between passing the constitution and creating the respective law. One could further compare the constitutionally mentioned maximum length of states of emergency with the length actually realized. Other potentially relevant aspects include: How many special courts have been established? What is the proportion of trials handled via such courts? Has the military been used domestically? Have actors involved in either the declaration or the implementation of a state of emergency been prosecuted after the dissolution of the state of emergency in case their behavior was not within the confines of legal action? Yet, any generally valid answer to all these questions must rely on some form of comparable operationalization of the features of emergency constitutions. As such, the findings in this article are arguably necessary to our ability to answer such questions in future research. We thank participants of the jubilee conference of the Eucken Institute (Freiburg, June 2014), the Public Choice Society Meeting (San Antonio, March 2015), the European Association of Law & Economics (Vienna, September 2015), and economic workshops at the Universities of Ankara (Bilkent), Bonn, Hamburg (Helmut Schmidt University), Lugano, Munich (TU), Prague (University of Economics), and Stellenbosch as well as Francisco Castellanos, Rola El Kabbani, Jerg Gutmann, Sven Höppner, Jean Michel Josselin, John Matsusaka, Stephan Michel, Konstantinos Pilpilidis, Agnes Strauss, Alois Stutzer, and Rok Spruk for valuable comments on earlier versions. All remaining errors are entirely ours. Footnotes 1 Zachary Elkins, Tom Ginsburg, & James Melton, The Endurance of National Constitutions (2009). 2 For the period between 1996 and 2004, Richards and Clay report thirty-five countries to have declared a state of emergency at least once, see David Richards & K. Chad Clay, An Umbrella with Holes: Respect for Non-Derogable Human Rights During Declared States of Emergency, 1996–2004, 13 Hum. Rts. Rev. 443 (2012). According to our search, we found a higher number, namely, forty-nine for the time period analyzed by Richards and Clay. Id. 3 Christian Davenport, Constitutional Promises and Repressive Reality: A Cross-National Time-Series Investigation of Why Political and Civil Liberties are Suppressed, 58(3) J. Pol. 627 (1996). 4 Linda Camp Keith & Steven C. Poe, Are Constitutional State of Emergency Clauses Effective? An Empirical Exploration, 26(4) Hum.Rts. Q. 1071 (2004). 5 Brian Loveman, The Constitution of Tyranny. Regimes of Exception in Spanish America (1993) is a very thorough analysis of the genesis of the emergency constitutions of many Latin American countries in the nineteenth century. 6 The text of the so-called constitution of the year three is documented at https://chnm.gmu.edu/revolution/d/450/. 7 All dates from Elkins et al., supra note 1. 8 It is, indeed, quite rare for emergency provisions to be added to a constitution by amendment. The only clear examples in democratic countries are Ireland in 1931 and Luxembourg in 2004; Germany provides a slightly different case as emergency provisions were added in 1968 to a constitution that had been devised while the country was still effectively occupied by Allied forces. A few other examples from non-democratic countries exist, in particular Comoros (2009), Cuba (1992), and the Soviet Union (1938). Bhutan is a special case in which emergency provisions were introduced in an interim constitution in 2005 that eventually carried over into the new constitution three years later. 9 Rafael La Porta, Florenzio Lopez-de-Silanes, & Andrei Shleifer, The Economic Consequences of Legal Origins, 46(2) J. Econ. Literature 285 (2008). 10 In addition to the countries already mentioned, these are El Salvador (1841), Haiti (1843), Mexico (1843), Dominican Republic (1844), Paraguay (1844), Guatemala (1845), Costa Rica (1848), and Honduras (1848). 11 Elkins et al., supra note 1. 12 The number of states also increased over this period. Expressed as a proportion, less than 60 percent of all states had an emergency constitution in 1850 whereas today, some 90 percent of all countries do. The source for our numbers is the Comparative Constitutions Project. The exact wording of the variable on which the numbers are based is: “Does the constitution have provisions for calling a state of emergency?” In each case where the question was answered in the affirmative, the country was coded as having an emergency constitution in the respective year. 13 According to the Comparative Constitutions Project, only seventeen countries have discarded constitutional emergency provisions. Elkins et al., supra note 1. 14 While an emergency constitution per se has not been re-introduced, the Austrian constitution has at least two clauses that could be called its emergency constitution: (i) art. 5(2) stipulates that in extraordinary times the president may issue decrees and (2) art. 79(2).2 that the military is to intervene as a consequence of natural disasters of extraordinary size. 15 The distinction between “new” and “amended” constitutions is not watertight. In our attempt to separate the two, we follow the classification of the Comparative Constitutions Project. Elkins et al., supra note 1. In the particular case of the Czech Republic, an amendment in 2009 allowed the parliament to dissolve itself under specific circumstances, which would include drastic emergencies. 16 Elkins et al., supra note 1. 17 These six potential justifications for declaring a state of emergency have been coded within the Comparative Constitutions Project. No further explanation on how to interpret them has been offered by those who created the dataset. We note that although some of these justifications are conceptually overlapping, we observe no emergency constitutions that, for example, mention both “general danger” and “threat to constitutional system” as separate items. 18 In principle, it is conceivable that states of emergencies are declared automatically after some pre-defined event occurred, such as an earthquake with a power of larger than 6 on the Richter scale. We are, however, not aware of any such automatism. 19 Throughout the article, we refer to this as separation of powers. Alternatively, one could also refer to the number of veto players, i.e., the number of actors whose consent is needed to make a specific decision. 20 An additional point that should ideally be included is who has the power to end a state of emergency, or who has the power to prolong an emergency. However, most emergency constitutions are surprisingly vague on this issue. We therefore refrain from including this aspect in the INEP, as the majority of the modern constitutions would have to be coded as having vague or no provisions. 21 Thorsten Beck et al., New Tools in Comparative Political Economy: The Database of Political Institutions, 15(1) World Bank Econ. Rev. 165 (2001). 22 Witold Jerzi Henisz, Political Constraint Index (POLCON) (2010). 23 Linda Camp Keith, Political Repression—Courts and the Law (2012). 24 According to the dataset from Linda Camp Keith which covers up to 181 countries. 25 Bruce Ackerman, The Emergency Constitution, 113(5) Yale L.J. 1029 (2004). 26 Keith,supra note 23. 27 John Ferejohn & Pasquale Pasquino, The Law of Exception: A Typology of Emergency Powers, 2 Int’l J. Const. L. 210, 216ff. (2004). 28 71 U.S. 2 (1866). 29 323 U.S. 214 (1944). 30 E.g., John Locke, The Work of John Locke in Nine Volumes. Volume 4: Economic Writings and Two Treatises of Government § 240 (1824). 31 Ackerman, supra note 25. 32 David Dyzenhaus, The Constitution of Law. Legality in a Time of Emergency (2006). 33 In the companion paper on consequences of emergency constitutions, we return to this issue by exploring the risk that emergencies lead to regime change, and in particular to loss of democratic institutions and violations of human rights. 34 Clinton L. Rossiter, Constitutional Dictatorship—Crisis Government in the Modern Democracies, with a new introduction by W.J. Quirk ch. IV (2009) gives an excellent account of that concept. It would be interesting to see whether the état de siège is followed more often by military regimes than other kinds of emergency concepts. Again, however, we delegate this question to a companion paper on consequences of emergency provisions. 35 This data is from the Comparative Constitutions Project again. 36 Keith,supra note 23. 37 In many cases, the CCP notes that central features of the emergency provisions are uncertain or unknown. We have rechecked many of these and only been able to fill out the blanks in the present German Constitution, the 1901 Danish Constitution, and the present constitution of Trinidad and Tobago. In all other cases we checked, we agreed with the CCP coders that not enough information was available to code the provisions. 38 The alternative would be to apply some form of multiplicative rule for how to aggregate the overall index. However, with a multiplicative index, one implicitly assumes that the single elements are complements such that, e.g., strict limits on declaration power reinforces the effects of having strict limits on the ability to suspend basic rights. We do not see any theoretical arguments for assuming that single elements are complements. Instead, our theoretical considerations lead us to believe that various provisions can substitute for each other, which necessitates applying and additive aggregation method. 39 Throughout, we code vague provisions as providing weaker emergency powers than explicit provisions providing some discretionary power. The power to derogate basic rights provides a pertinent example that we often observe in emergency constitutions. When the emergency constitution does allow some rights to be suspended, most provide an explicit list. When the constitution is vague on this issue, it does not contain a list. We therefore code vague provisions as a weaker provision than when a limited list is provided because the vague provisions may mean that some rights may be suspended—similarly to providing a limited list—and it may mean that they may not. 40 The dataset by Keith covers the period from 1979 until 2009 (Keith,supra note 23). During the first two thirds of this period, there was a slight increase in the separation of powers under states of emergency. However, this tendency is dominated by a change in the opposite direction over the entire period here under consideration, and particularly from 1950 until 1975 when decolonization was effectively through. 41 The overall trend describes an average of countries introducing substantially stronger emergency powers and countries weakening these powers. However, even when exploring the simple changes in the INEP, it is evident that about double the number of countries have increased rather than decreased the index while about half of the sample has had unchanged emergency constitutions since either 1950 or independence. 42 José Antonio Cheibub, Jennifer Gandhi, & James Raymond Vreeland, Democracy and Dictatorship Revisited, 143 (1–2) Pub. Choice 67 (2010). 43 E.g., Rossiter, supra note 34 (in particular ch. X). 44 Id. at 141. 45 Joseph F. Hair, Jr., et al., Multivariate Data Analysis (1998). 46 Countries without emergency constitutions are left out of this assignment. 47 A possible objection to analyzing both valid and defunct constitutions simultaneously could be that there might be some general trends that lead twenty-first-century emergency constitutions to be very different from nineteenth-century constitutions. If we reduce the analysis to only those constitutions currently in place, we lose around two-thirds of all our observations. Yet, clusters 1, 3, 5, and 6 remain stable whereas clusters 2 and 4 shift around a bit. 48 Which seems certainly in line with the fact that around one-third of them have been written by governments that came to power via a coup d’etat. 49 Supra note 42. 50 Supra note 22. 51 Angus Maddison, Statistics on World Population, GDP and Per Capita GDP, 1-2008 AD. Online database, available athttp://www.ggdc.net/MADDISON/oriindex.htm. 52 Our full dataset includes 351 constitutions, leaving 67 constitutions from countries with only one emergency constitution in the data. However, it must be noted that most of these countries have had more than one constitution although only one appears in the current data. The full CCP dataset, for example, includes information on four of Denmark’s five constitutions since 1849, but only the current 1953 constitution includes emergency provisions. 53 Elkins et al., supra note 1. 54 Christian Bjørnskov & Stefan Voigt, The Determinants of Emergency Constitutions (mimeo) (2016). 55 Christian Bjørnskov & Stefan Voigt, Why Do Governments Call a State of Emergency? On the Determinants of Using Emergency Constitutions, Eur. J. Pol. Econ. (forthcoming 2018). © The Author(s) 2018. Oxford University Press and New York University School of Law. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

International Journal of Constitutional LawOxford University Press

Published: May 12, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off