Emergency powers in democratic states: Introducing the Democratic Emergency Powers dataset

Abstract

This article discusses the content and causes of emergency powers in democratic states. It makes two main contributions. First, it introduces the Democratic Emergency Powers dataset, which codifies formal rules of institutional change during periods of emergency for all democracies from 1800 to 2012. These data are important because often state institutions in times of emergency differ dramatically from those in times of relative peace, and this influences government decision-making during times of crisis. I discuss the collection process and develop a measure of the latent strength in each of these emergency provisions. Second, this article examines the determinants of emergency power strength in democratic nations cross nationally. I find that previous and current conflict experience influence emergency power provision and strength, but also that domestic determinants have an impact on the types of instruments states institutionalize for times of emergency.

Introduction

In the past decade, states of emergency have been declared in Ecuador (2010), France (2015), Belgium (2015), Turkey (2016), the Philippines (2016), and the United States (2019), in response to terrorist violence, coup attempts, and domestic disturbances. These seemingly undemocratic events find homes in the laws of many democratic states, as from 1800 to 2012, 80 democracies—over half of all democratic country-years during this period—formally instituted “emergency power” provisions specifying a shift in the institutional structure of the nation during times of crisis.

Examining each of these cases demonstrates how the effects of these emergency powers can be quite dramatic and long lasting, particularly the frequency of their abuse (Rooney, 2019). However, outside early theoretical work (Rossiter, 1948), we know very little about the causes and consequences of these powers. In this article I introduce the Democratic Emergency Powers (DEP) dataset, which codifies the formal rules regarding states of emergency in democracies across time, including the circumstances in which states of emergency can be declared, how long they last, the state’s institutional structure during the state of emergency, and the extent and limitations of the powers granted to the state’s leader to combat the crisis. Using Bayesian factor analysis, I develop a measure of the latent strength of the power of the executive in each emergency provision.

I then empirically examine the origins of these powers. Elements that lead to an expectation of future turmoil, such as a history of violent political conflict or a looming international threat, increase the likelihood that states adopt stronger emergency powers. Additionally, states enact stronger emergency provision when they posses elements that exacerbate a time of emergency, such as features of the state that could increase grievance or decrease the government’s control during times of crisis. These data allow the empirical examination of emergency provisions in democratic states, and provide the first quantitative analysis of their causes.

A database of emergency provisions

Emergency provisions are legal acts that empower the government to take actions beyond its standard procedures in times of grave crisis, such as an external conflict, domestic rebellion, or natural disaster. Tracing their origins to the Roman Republic (Rossiter, 1948), emergency provisions are common institutional features of democratic states with a variety of constitutional arrangements, and have been adopted consistently with the spread of democracy. Each provision is designed to overcome restraints on necessary action during periods of national distress. However, emergency provisions widely vary in terms of the power that states grant to the executive to combat a period of crisis.

Each emergency provision highlights the extraordinary nature of the threat to the democratic order and national survival. Yet how states choose to combat this threat is often very different. This variation is evident in the required cause of an emergency, ranging from armed aggression against the nation (Moldova, 1994) to any apparent threat of public danger (Serbia, 2006). It is even more prominent in the range of powers granted to the leader of the nation, where some provisions do little to shift the balance of power between consulting government actors such as the President and Prime Minister (Portugal, 1976), and others grant the state leader the power to act above all others to do what is “strictly necessary to repel the aggression” (Columbia, 1992, Article 212). To understand the importance of emergency powers, it is therefore vital to capture this variation in strength.

To this end, I have developed the DEP dataset. The data cover all states scoring 6 or higher on Polity’s composite Polity2 score for each year from 1800 to 2012 (Marshall and Jaggers, 2001). Under this criteria, 113 states are democratic for at least 1 year, from which there are 147 individual constitutions and over 500 potentially relevant amendments and legislative acts for democratic states. Examination of these documents provides 172 unique emergency power provisions. For each provision, I first apply a simple coding procedure to categorize each emergency provision according to their dominant sources of variation. I first assess when a formal state of emergency can be declared according to each document. This includes first whether the democracy legally provides for a state of exception and what government actors or bodies can formally declare the state of emergency. I then code a series of indicators of the specific circumstances—such as external attack, war, foreign aggression, civil conflict and upheaval, and natural disaster—under which actors may declare a state of emergency.

I document changes to the traditional institutional order during a period of emergency, coding a series of indicators of the level of formal power diverted to the effective leader in times of emergency, using information on the effective leader for each state from Archigos (Goemans, Gleditsch and Chiozza, 2009). These measures of formal power indicate whether the power to declare an emergency belongs to the effective leader or some other political actor, or requires consultation between the leader and outside actors. I then code whether the effective leader can make use of additional powers in times of emergency and assessed the nature of these powers, regarding clarity, whether they allow the leader to take measures outside of a military scope, and whether they grant the ability to make policy declarations. Lastly, I code measures that indicate the number of prospective and retrospective checks on actions taken during an emergency, as well as formal limitations on how long policies declared during an emergency and emergencies themselves may last.

This provides a series of binary, ordinal, and count measures coded such that higher values indicate greater empowerment for the leader during an emergency, To convert these factors into a measure for statistical analysis, I use Bayesian factor analysis for mixed data to derive a measure of latent emergency power strength for each document using Markov Chain Monte Carlo methods. ¹ In factor analytic models, the observed measures corresponding to the elements within constitutional provisions are assumed to be intercorrelated because they are driven by the same latent construct—emergency empowerment. ² Each dimension is related to the strength of the powers granted to the executive in times of emergency, but it is not obvious to what extent they contribute to strength. The measurement model leverages the variation in indicators across each individual case to model both the relationship between each indicator and the latent construct and the resultant estimate of the latent strength for each emergency power provision (Benson and Clinton, 2014). Rather than focusing on how each indicator contributes to emergency power strength, I note only that they validate expectations, and focus on the individual estimates. ³ The procedure provides a continuous measure of emergency power strength for each emergency power provision in each year, ranging from -1.02 to 1.64, with mean 0 and a standard deviation of 1. ⁴

Of the 172 legal documents in the sample, 117 begin during a democratic year, whereas 55 are holdovers from a period of non-democracy. Much of the variation in these data is cross-national (between standard deviation: 0.80). Indeed, 65% of democracies have stable emergency provisions over time, with 22% seeing one change, and only 13% having more than two emergency provisions. However, such change can be quite dramatic (within standard deviation: 0.52). To provide one example, emergency power strength in Sri Lanka ranges from -1.02 in 1957 to 1 in 1981, covering nearly the entirety of the range of emergency power strength.

Emergency powers in 1989 and 2012

To further demonstrate the variation present in the data, in Figure 1, I show the range of constitutional emergency powers for democratic states for 1989, at the close of the Cold War. ⁵ There are several descriptive takeaways from these data. First, states with stronger emergency powers include those such as France and Turkey who have recently experienced states of emergency. Others with strong emergency powers, such as Trinidad and Tobago and Costa Rica, suggest it is not simply the most obvious aspects of the international or domestic environment, such as military prominence or democratic instability, that drives emergency power provision. Lastly, the strongest emergency powers in 1989 are from the Republic of Korea, whereas the weakest emergency powers in 1989 are from Japan. This is noteworthy because the states face similar international environments and the constitution of the Republic of Korea borrowed extensively from the constitution of Japan.

Figure 1.

Constitutional emergency power strength across countries: 1989.

In Figure 2, I show the range of emergency powers in 2012. Owing to the rise of democracy in the post-Cold War era, the number of states with emergency powers has risen considerably in this time, primarily in Central and Eastern Europe and Africa. Further, this increase in emergency power provision includes strength, both high (Sierra Leone) and low (Ukraine). There are also changes across these two time periods within states, with states both expanding and contracting their emergency powers. In just the roughly two decades considered here, Argentina increased the strength of its emergency powers, and Bolivia reduced the strength of its powers. These figures demonstrate there is significant variation in strength across states that cannot easily be explained as indicative of alternative state characteristics.

Figure 2.

Constitutional emergency power strength across countries: 2012.

I further find evidence that the measure does not reflect previous conceptions or data on executive empowerment and democracy. Although Polity scores are designed to measure peacetime democratic institutions, scholars frequently include them in models of international or domestic conflict, when emergency provisions are likely to be in effect. When scholars use Polity scores to differentiate between democracies and autocracies in models of conflict, their implicit assumption is that the institutions captured in the scores are static with regards to crisis or that states with stronger democratic institutions would also be more democratic in times of conflict. To challenge these assumptions, I examine the relationship between emergency power strength and Polity scores. This is a particularly apt comparison, because scholars argue Polity scores primarily reflect the decisional constraints on the executive (Gleditsch and Ward, 1997; Bennett, 2006). I present a scatterplot of the relationship in Figure 3, comparing the individual emergency power scores for each country year to each country’s Polity score in that year. If Polity scores capture democratic constraints on the executive during times of emergency, states with lower Polity scores should be more likely to grant strong emergency powers in times of crisis. Instead, there is almost no relationship between emergency power strength and strength of democracy as measured by Polity.

Figure 3.

Emergency power strength and polity scores.

This description of the data reveals there is substantial variation in executive empowerment that current data do not capture. As the motivating examples show, the effects of these emergency powers can be quite dramatic and longlasting. Although some have explored how these emergency powers may be important in today’s changing world (Ferejohn and Pasquino, 2004; Rooney, 2019), our understanding of the origins of these powers is limited. In the following section, I explore the theoretical and historical explanations for the origins of emergency powers to provide a basis for testing their influence and to evaluate the validity of the data.

Emergency power origins

Although there are few direct accounts of the decision to adopt emergency powers, historical accounts of constitution-making in democracies suggest elements that influence the decision to adopt emergency powers. In this section, I explore these determinants in greater depth for democratic states from 1950–2012 using the DEP data. I model each of the predictors first as determinants of legal changes to the constitutional order and then as a determinant of emergency power provision using a Heckman selection model. I explore first the inclusion of emergency powers, for which I code all states that install formal emergency powers as 1 and all states with no formal emergency powers as 0. Second, I explore emergency power strength, where I use the full emergency power strength variable.

The first determinant of executive emergency power is the strength of the executive in times of peace. Leaders that face particularly strong domestic constraints on their actions during peacetime require greater freedom of action to overcome emergency situations, especially in the case of states with a strong separation of powers and complex governments (Rossiter, 1948). Emergency powers historically result from the desire to break deadlock, and thus systems with a greater potential for institutional gridlock should be likely to prescribe stronger emergency powers. I include two measures of institutional arrangements during peacetime. I first account for the possibility that a greater number of veto points in a system leads framers to institute stronger emergency powers using POLCONIII (Henisz and Zelner, 2002). Because the balance of power in government at the time of constitutional reform might likewise influence the provision of emergency powers (Negretto, 2013), I include a measure of legislative fractionalization (Henisz and Zelner, 2002).

Framers often grant emergency powers when the nation’s standard democratic institutions have proven unable to combat domestic discord. In drafting the Venezuelan Constitution of 1961, constitutional framers, like those many in nascent democracies, recalled memories of the nation’s previous experience with autocracy and its long history of domestic coups and military intervention in the government (Manrique, 1988). I test for how the quality of a state’s domestic political institutions reflects the probability of the breakdown of the normal government order in three ways. First, because democratic constitutions often specify emergency action to battle coup d’états and civil discord, I include the count of previous coup d’états in a state (Powell and Thyne, 2011). Second, states with stronger democratic institutions may be less vulnerable to internal or external conflict. As a result, I include each state’s Polity score to account for the vulnerability of peace time institutions to crisis (Marshall and Jaggers, 2001). Lastly, states that face persistent change in domestic institutions may be particularly unstable and unable to combat emergency situations. As a result, I include a measure of the occurrence of a regime change (a three-point change in Polity score), in the previous 3 years.

Framers of democratic institutions also grant emergency powers to the executive due to the high stakes of international conflict. States may consider stronger emergency powers a necessity when previous conflict drives expectations of future conflict, or because a culture of conflict drives both past and future disputes. Spurred on by the memory of World War I and World War II, the constitutional framers in France in 1958 believed the constraints on the executive had undermined his ability to enact necessary policies in the face of an existential threat (Foyer, 1988). France’s constitutional process thereby granted immense power to the president in times of emergency to ensure such calamity would not recur. I measure conflict as the count of previous militarized interstate disputes involving the state up to the given year (Sarkees and Wayman, 2010).

States may also consider adopting emergency powers in response to a particular threat. States often engage in interstate rivalries, continuous conflicts with the same opponent in which each conflict makes makes future conflict more likely (Colaresi and Thompson, 2002). Such rivalries lead to the rational expectation of future disputes, and the state may thus be more likely to enact emergency powers. Greece’s long-held rivalry with Turkey over Cyprus demonstrated the government’s consistent inability to gain the upper hand on their opponent (Tsatsos, 1988). Framers of the Greek Constitution of 1975 therefore sought strongly to enact provisions to ensure future governments would have the power necessary to prosecute future disputes with Turkey. I therefore include a measure for the presence of an interstate rivalry (Thompson and Dreyer, 2011).

States should be equally concerned about factors that might make future internal discord more likely or dangerous. States with larger populations may fear they will be unable to control the population in an emergency, and provide greater power to the executive for such contingencies (Collier and Hoeffler, 2002). I include a measure of the natural log of the state’s population (World Bank, 2007). States may craft emergency power provisions based on their expectations of future financial burdens or collapse, which can exacerbate periods of crisis or may themselves be emergencies. To account for this, I include a measure of the health of the state’s economy, as captured by the natural log of the democracy’s gross domestic product (GDP) per capita (Gleditsch, 2002). I also include a measure of material capabilities using the Composite Index of National Capability (CINC) scores (Singer, Bremer and Stuckey, 1972), because states with stronger material capabilities should have more resources to combat or prevent emergencies (Fearon and Laitin, 2003).

I present all results using state-clustered standard errors and include cubic splines for the number of years the state has been democratic to account for democratic institutional duration dependence, and for the number of years since constitutional change, as predictors of formal legal changes. Table 1 displays the results for emergency powers, whereas the results for legal change are included in the online appendix.

Table 1.

Determinants of new emergency powers.

	Emergency power provision	Emergency power strength
	(1)	(2)
Past MIDs	0.038*	−0.005
	(0.022)	(0.006)
Rivalry	−0.818	−0.071
	(0.719)	(0.243)
Regime change in past 3 years	0.806*	0.089
	(0.420)	(0.238)
Previous coups	0.420***	0.124***
	(0.148)	(0.041)
Polity score	−0.027	0.111
	(0.227)	(0.087)
Legislative fractionalization	−3.492*	−1.312**
	(1.928)	(0.669)
Veto players	0.725	−0.164
	(2.151)	(0.889)
Log of population	−0.325	0.232**
	(0.257)	(0.092)
Log of GDP per capita	0.009	0.049
	(0.355)	(0.126)
CINC score	179.697**	−4.514
	(84.301)	(13.265)
Constant	3.256	−3.004**
	(3.488)	(1.224)
N	2299	2299

Clustered standard errors appear in parentheses below the estimated coefficients of all models.

*

$p<0.10$ , ^** $p<0.05$ , ^*** $p<.01$ .

Looking at Model 1, I first note that 60% of the changes to the constitutional order include the provision of emergency powers, whereas 40% do not. I find that previous international conflicts are a strong predictor of emergency power provision, whereas rivalry exerts no impact on emergency power provision. This suggests it is not the external rival that drives emergency power adoption, but rather the expectation of future conflict driven by experience.

Previous internal conflict is a substantively and statistically strong predictor of emergency power adoption. Combined with the positive and significant effect of regime change, this shows that states with significant or recent experience with internal discord are likely to provide for future crises. Political institutions in times of peace demonstrate a mixed relationship. Although political institutions captured by polity and veto points demonstrate no relationship, legislative fractionalization makes adopting emergency provisions less likely, perhaps because the fractured groups are unwilling to grant strong power to a single individual. Population size and GDP per capita exhibit no relationship with emergency power provision. Material capabilities, however, are a strong predictor of emergency power adoption.

Model 2 captures the predictors of emergency power strength. Whereas several of the substantive conclusions remain the same, others change dramatically. Previous coups remain a strong predictor of emergency power behavior. However, although previous militarized interstate disputes were a strong predictor of emergency power adoption, they exhibit no relationship with emergency power strength. Instability as marked by a regime transition in the previous three years, although fundamental to the adoption of emergency powers, does not make emergency powers stronger. Although material capabilities do not predict emergency power strength, the size of the state’s population does, as larger states may require greater tools to handle a period of extreme crisis. The only other variable that reaches traditional levels of significance is legislative fractionalization, which again has a negative effect on emergency power provision.

Conclusion

In sum, this project provides several important contributions to the study of international relations and comparative politics. The primary contribution is the systematic collection of emergency provisions in democracies across time. I codify the emergency provisions in the constitutions and legislation of 113 democracies from 1800 to 2012, including all changes or amendments, and measure emergency power strength. Additionally, I demonstrate how state level factors contribute to the decision to grant emergency powers and find that the determinants of emergency power provision and emergency power strength are only partially in agreement. A history of external conflict predicts emergency power provision, but expectations about internal turmoil seem to predict the strength of these powers to a much larger degree. These data provide an important source of comparative institutional variation for democratic states that recent history suggests is increasing in importance. Further, this institutional variation may have an impact on any number of foreign or domestic policy outcomes, such as human rights, counter terror, disaster preparedness, and domestic political competition. The possibilities stretch across many areas of study and encompass a variety of potential methodologies.

Supplemental Material

Appendix – Supplemental material for Emergency powers in democratic states: Introducing the Democratic Emergency Powers dataset

Supplemental material, Appendix for Emergency powers in democratic states: Introducing the Democratic Emergency Powers dataset by Bryan Rooney in Research & Politics