Abstract
Martial law is widely used by governments around the world and yet remains poorly understood. In this article, I aim to expand what we know about the concept of martial law, within the context of the broader literature on emergency powers. This article begins by offering an academic definition of martial law as well as a measurement strategy, generating cross-national, time-series data. This article introduces a novel dataset that records declarations of martial law and states of emergency from 1995 to 2022. I illustrate the utility of this data by showing it can complement research on government repression and overreach.
Introduction
Just 10 years after a decades-long labor to build durable democratic institutions, the military in Myanmar staged a coup and ousted leader Aung San Suu Kyi. After the 2021 coup, the newly installed military junta in Myanmar began to attack perceived opponents. Leadership began declaring martial law in different regions of Myanmar, beginning with parts of Yangon and continuing into February 2023. The combined effect of martial law and targeting medical personnel has led to grave human rights abuses, including denying critical life-saving medical care for infants (Gelineau et al., 2021). In 1982, Amnesty International issued a report with interviews of people experiencing the effects of martial law in the Philippines, which had been declared in 1972 by then-President Marcos. In the report, one Filipina woman describes her treatment at a “safehouse” in this case, another word for detention center, where she was subjected to repeated torture and sexual abuse. She was denied legal counsel, medical care, or any communication with her relatives. The report goes on to detail that nearly 50,000 people were detained in the first three years of the order, many without a trial (Amnesty International, 1982). In 2022, Human Rights Watch reported that President Felix Tshisekedi of the Democratic Republic of the Congo had declared martial law in North Kivu and Ituri, provinces in the Eastern part of the country, a year earlier. Since then, military personnel have regularly used lethal force, conducted searches of citizens’ homes, and arrested anyone deemed disorderly (Human Rights Watch, 2022). These examples are but a few from the universe of martial law, which states can rely on to respond to a variety of circumstances. There are many more examples than those provided here.
A surprising quality of martial law is that despite its frequent discussion in public discourse, journalism, and in other spaces, there is no standard academic understanding of what martial law is nor a measure of its specific use around the world. It is not hard to imagine why; martial law differs considerably between countries. Some governments have used martial law as a blanket order across the entire country, such as the Philippines in the 1970s or Pakistan at various points across the last five decades. However, many other governments target martial law orders to specific districts, such as in Turkey or India. Some governments have martial law orders that span years, while others expire in weeks. We know much more about emergency powers in broad terms than martial law as a standalone concept. This growing literature represents an opportunity to build out a specific definition of martial law as a distinct concept, situated in the overall topic of emergency powers. Thus, this article proposes a definition of martial law and introduces a cross-national dataset cataloging its use.
Martial law represents one of many ways to respond to an emergency, particularly a security threat, and examining its use can germinate a new topic of discussion for international security scholars. An academic standard for the use of martial law specifically, rather than emergency powers or states of emergency broadly, will allow scholars and policymakers alike to better understand its consequences, when states tend to use it, and most importantly, when it is abused. It may seem obvious to state, but this discussion matters because martial law can introduce conditions for grave human rights abuses and create incentives for executives to expand their power. Defining and measuring martial law separately and specifically allows scholars to pose research questions such as how often martial law orders are declared in anticipation of violence or after violence has already occurred, what sort of regimes are more likely to use martial law as opposed to a state of emergency, and if and how declarations have changed over time. The following analysis therefore adds another perspective to the ongoing discussion on the topic.
I define martial law as an explicit and legally justified declaration by the central or regional government allowing the temporary substitution of local rule by the military. The definition makes explicit what a debate in the literature regarding emergency powers in the early 2000s only danced around (Ferejohn and Pasquino, 2004; Scheuerman, 2006). From Roman constitutions to John Locke, to George W. Bush's decision-making in the immediate post-9/11 world, legal scholars have argued over where emergency powers should be situated when ”squaring the circle” between upholding rule of law and prioritizing emergency response (Scheuerman, 2006). This article reactivates this discussion by zooming in on martial law in the broader context of emergency powers and introducing a novel dataset. As rulers regularly use emergency powers to respond to natural disasters, disease, violence and unrest, and other crises, there are ample opportunities to put the use of martial law into sharp focus.
I collect new data on declarations of martial law and states of emergency from existing international news sources and introduce it with this paper. The empirical dataset records declarations of states of emergency and martial law, as well as where these declarations were directed. It includes data from 172 countries over 27 years, from 1995 to 2022. Over those years, 156 unique countries declared martial law or a state of emergency at least once, leading to a total of 2142 declarations. Of those declarations, 82 were explicitly martial law. A brief application of these data shows that active rebel groups are a strong predictor of emergency declarations for violence and unrest, as well as a weak but positive correlation with human rights abuses.
In the next sections, I review existing literature on emergency powers, introduce a formal definition, and describe the measurement and data collection strategy.
Literature review
Although martial law is often invoked in moments of extreme crisis, the specific concept of martial law is undertheorized and remains empirically elusive in political science. 1 This omission is due to both limited data availability aND conceptual ambiguity. Rather than existing as a clearly defined subject of inquiry, martial law appears at the margins of literatures on repression, constitutional design, civil conflict, and perhaps most relevant, emergency powers. In order to clarify this ambiguity, this article grounds the concept of martial law within the broader concept of emergency powers, where emergency powers exists in broader theoretical tradition of the “state of exception.” In the “state of exception,” constitutional norms are suspended under the legal pretense of preserving political and physical order. Within this framework, martial law does not represent an illegal seizure of absolute power, rather it is a formally declared, territorially and temporally bounded expansion of executive authority that substitutes military for civilian governance in response to perceived threats.
The foundational concept of the state of exception, drawn from the work of Schmitt (1921, 2005) and elaborated by Ferejohn and Pasquino (2004), helps clarify when governments take extraordinary measures to respond to emergencies, and therefore intrduces the role of emergency powers. Ferejohn and Pasquino (2004) argue that modern constitutional orders rest on a latent duality: a normal legal order for peacetime governance and an exceptional legal order for moments of existential crisis. These moments of suspension are not external to law, but are instead authorized by it. Scheuerman (2006) extends this logic by outlining several normative perspectives on emergency rule, among them constitutional relativism, extralegalism, and legal formalism. He argues that legal formalism, which insists on both ex ante constraints and ex post accountability, provides the only normative foundation capable of taming the inherently destabilizing logic of the exception. Yet even in Scheuerman's analysis, martial law is not identified as a distinct subtype. Rather, it is subsumed into a broader category of emergency powers, where internal structure remains murky. This represents an opportunity to contribute more specificity and conceptual clarity.
Emergency powers, states of exception, and repression are all closely related to martial law, yet none are conceptually identical. To clarify how martial law fits into this family, it is helpful to begin with the overarching idea of emergency powers as instruments that enable a government to legally transcend constitutional limits in moments of crisis. These powers are typically established in constitutional clauses or enabling legislation that define when, by whom, and for how long such authority may be invoked. As Silverstein (2023) notes, emergency powers can originate in the oldest traditions of legal governance, stretching back to the Roman dictatorship, which allowed temporary concentration of power in the hands of a single magistrate to preserve the republic.
Recent research has focused primarily on the broader category of emergency powers. Rooney (2019) demonstrates that democracies with constitutions granting strong emergency authority tend to experience greater conflict propensity, suggesting that sweeping executive leeway can provoke instability rather than resolve it. Lührmann and Rooney (2021) find that states of emergency are strongly associated with autocratization. Bjørnskov and Voigt (2021, 2024) contribute important evidence on the correlates of emergency powers declarations and regime type, showing that disasters and terrorist events frequently lead to emergency declarations, and that weak constraints on executive authority tend to correlate with more deaths, more severe repression, and longer-lasting crises. Ginsburg and Versteeg (2021), in contrast, emphasize institutional resilience, showing that legislatures and subnational actors provided important checks on emergency governance during the COVID-19 pandemic. Appel and Smith (2025) show that natural disasters and repression are not mutually exclusive; rainfall shocks increase state incentives to engage in human rights violations, as governments respond to food scarcity with coercive strategies to maintain control.
There is also work in the civil conflict space. Aydin and Gray (2023) argue that while emergency rule grants counterinsurgents broader powers to suppress insurgencies, such measures can backfire: delaying emergency declarations signals desperation, erodes civilian support, and hinders efforts to win “hearts and minds.” Therefore, the timing and expansion of constitutional emergency powers can reduce civilian backing for those in power, undermining counterinsurgency outcomes. Further, detentions of activists under emergency rule can mobilize non-violent civilian opposition and shift electoral loyalties toward an ongoing insurgency's political allies, showing that repressive tools like emergency powers in the context of prolonged civil war can backfire (Aydin and Emrence, 2021).
Yet emergency powers do not automatically entail martial law. In the Ferejohn and Pasquino (2004) concept of the state of exception, the authors distinguish between the formal legal order and its suspension. In their formulation, the exception is legally recognized even as it suspends legality, creating a space in which the law authorizes its own interruption. Martial law operates within this paradox: it is declared as a lawful response to threat, but it can authorize actions that would otherwise be impermissible and unthinkable, such as military governance, suspension of civil liberties, and trial by tribunal.
The absence of martial law as a specific concept is not unique to legal theory. In political science, empirical work on emergency rule has rapidly expanded in recent years, but continues to gloss over martial law as a specific institutional form. Early work by Davenport (1996) explored constitutional emergency powers and their effects on repression, finding that explicit legal guidance on emergency rule can reduce executive overreach. Constitutions that clarify when and how states of emergency may be declared tend to preserve more civil liberties than those that do not. Davenport's later work emphasizes how ambiguity in legal authority can lead to the expansion of repression under the guise of crisis governance (Davenport, 2007). Still, neither of these studies distinguishes martial law from other forms of emergency response.
In general, emergency powers are form of repression, but the specific repressive tactics come with a legal justification. Emergency powers reveal how legality can legitimize repressive state coercion, and scholars of state repression have examined some of the conditions under which governments use force against domestic populations. Indeed, repression is a strategic move on the part of elites. Oztig and Karluk (2025) determine that political elites use repression to leverage better diplomatic outcomes. They show that leaders strategically increase repression on internal Uyghur populations to maintain positive relations with China, which aligns with broader theories of emergency governance, where states justify extraordinary measures, including martial law-type interventions, by framing specific groups as existential risks. Davenport (2007) defines repression as “the actual or threatened use of physical sanctions against an individual or organization, within the territorial jurisdiction of the state, for the purpose of imposing a cost on the target as well as deterring specific activities and/or beliefs perceived to be challenging to government personnel, practices or institutions” (Davenport, 2007; Goldstein, 1978, p. 2). This wide-ranging definition includes martial law but does not distinguish it from more routine or covert forms of state coercion. Martial law may use repressive tactics, but repression itself does not necessarily mean martial law or military governance. However, repression also echoes themes in the literature on state responses to violence, including Kalyvas (2006). In his chapter outlining when indiscriminate violence occurs, Kalyvas describes military escalations within the state without explicitly naming martial law. He states that “Incumbent indiscriminate violence usually takes place in the context of military operations known as ‘mopping up,’ ‘comb,’ ‘cordon and search,’ ‘search and destroy’ or ‘scorched earth’ campaigns that seek to encircle and liquidate insurgents and undercut an insurgency's civilian basis. These campaigns are often dubbed ‘pacification’ campaigns” (149). Yet, while these tactics can resemble martial law in practice, they are not identified as martial law, nor are they tied to any formal declaration of exceptional status. They are practices of exception without the legal ritual of declaring the exception. Repression may be covert or overt, legal or extralegal, and does not require public declaration or institutional procedure. Martial law, in contrast, must be declared and justified, usually with reference to some constitutional or legal authority. It can therefore be thought of as a form of declared repression, distinct in its legal visibility and its aspiration to normalcy, however contested.
Consider some differences between Afghanistan and Burkina Faso. After the United States pulled troops out of Afghanistan in 2021, the Taliban quickly re-established control in Kabul for the first time since late 2001. Since August 2021, women's access to education, autonomy, and expression have been continuously and persistently curtailed (Faiez, 2023). In Burkina Faso, in June 2021, the central government extended a state of emergency in six distinct regions as a response to increasing violence from local armed groups (BBC Monitoring Africa, 2021). Both situations involved a government making unmistakeable attempts to repress specific groups of people. However, taken at face value, the actions of the Taliban government are obviously far different from the actions of the government of Burkina Faso. The Taliban's actions constitute ideological, gender-focused repression, whereas Burkina Faso's governmental actions were direct responses to internal conflict.
Despite the growing body of empirical work on emergency powers, and the long history of work on repression, martial law remains largely unexamined in its own right. One reason for this omission is conceptual collapse. Martial law is frequently treated as either directly synonymous with repression or as a vague extension of emergency rule. This paper argues that a distinction here is crucial. To study martial law as an institution requires attention to the formal, legal dimension of emergency governance. Where much of the repression literature treats coercion as either covert or informal, martial law is inherently declarative; it is a public, legally justified suspension of civilian rule. Moreover, I argue that martial law is also not synonymous with the broader state of emergency. The two may possibly and can often overlap, but they are not conceptually identical. For example, in the Philippines, President Duterte declared a state of emergency following the 2016 Davao bombing and then escalated to a formal declaration of martial law in Mindanao after increased violence by the rebel group Abu Sayyaf (Mapping Militant Organizations, 2022). The two events were legally and temporally distinct, with the latter involving the substitution of military for civilian authority in the affected region. This example illustrates the value of conceptual distinction: martial law is best understood as one specific instrument within the broader architecture of states of exception and emergency powers.
While states of emergency and martial law are often conflated, they differ in important ways. A state of emergency is typically declared under constitutional or statutory provisions that expand the executive's authority while keeping civilian institutions in place. Civil liberties may be curtailed, and the military may be deployed, but with legislative and judicial oversight often remaining intact. In contrast, martial law entails the substitution of military authority for civilian governance. Under martial law, the military may displace elected officials, replace civilian courts with military tribunals, and thoroughly suspend the civilian constitution. Indeed, with the novel dataset introduced in this paper, scholars can continue to hone the specific differences between martial law and states of emergency in practice.
The contribution of this paper lies in better defining, measuring, and recording the tool of martial law. Much of the existing research has focused on the constitutional provisions and not necessarily the specific events involving the use of these emergency provisions, in part because of data limitations. Lührmann and Rooney (2021) specifically update a previous dataset from Hafner-Burton et al. (2011) documenting states of emergency from reports from the State Department. However, these data do not draw distinctions for where the states of emergency are aimed and risks problems of ecological inference by regarding the entire country as under a state of emergency. A recent book by Bjørnskov and Voigt (2024) explores the vast universe of states of emergency, their constitutional derivations, and finally, their catalysts and impacts. However, the analysis relies upon an updated version of the Hafner-Burton et al. (2011) data. I contribute to this literature by diving deeper into international and local news sources rather than State Department reports, by documenting whether a country declared a state of emergency or martial law, by recording where this declaration was targeted, and finally, by documenting for what reason the emergency was declared, offering a sharper lens through which to examine the practice of emergency governance.
Definition
Scholars have drawn important distinctions between normal legal procedures and the extraordinary measures taken during emergencies, but the concept of martial law has received less focused attention. This section aims to fill that gap by offering a definition of martial law as a specific form of emergency rule within the broader framework of states of exception. Rooney (2019) defines emergency provisions as “legal acts that expand the purview of state actors during times of international or domestic crisis” (647). Lührmann and Rooney (2021) define a state of emergency as “the mechanism by which, in times of imminent danger, a government is empowered to take actions beyond its standard procedures” (619). However, previous definitions of the concept fall short in fully capturing the nuances and complexities of martial law events. These descriptions offer insight, but the ambiguity inherent in these definitions represent an opportunity to clarify further and define how governments use martial law as a subset of emergency powers.
This section will offer the following definition of martial law: martial law is an explicit and legally justified declaration by the central or regional government allowing the temporary substitution of local rule by the military. The section below will build the definition clause by clause.
An idealized example might look like military personnel “boots on the ground, so to speak,” arriving in a town on rumors that a rebel group is planning an attack or has already attacked. Troops impose curfews, replace local police, and work as rule enforcers. Traditional courts are replaced with military tribunals, and justice through civilian rule such as juries of peers is no longer available. Civilians experience random detentions with or without cause, on suspicion of some collaboration with the threat that motivated martial law in the first place, and possibly as a result from denunciations from neighbors or friends. Worse yet, executions based on real or perceived dissent of the government are common. This may seem a stylized and dramatic rendering of some fictional scenario, but it accurately describes the White Terror period in Taiwan, which lasted almost 40 years from 1949 to 1987.
Working backwards, the clause “temporary substitution of local rule by the military” takes root from EW Killam's 1989 definition of martial law. Killam (1989) states that “[martial] law involves the temporary substitution of military authority for civilian rule and is usually invoked in time of war, rebellion, or natural disaster” (annotation, n.p.). He first defined it in an American context and stated that it had been declared nine times since World War II in the United States. The definition is broad enough to carry forward to other countries and contexts. While Killam's definition offers a jumping off point, it can be refined for a more specific and modern understanding, to be better utilized in the social sciences.
Notably, the definition of martial law proposed in this paper uses the word “local” rather than “civilian.” A definition that is limited to replacing civilian rule makes states ruled by military dictators and juntas difficult to categorize. In practice, countries ruled by the military can and do declare martial law; as referenced in the introduction, Myanmar's military junta declared martial law in 2021. Moreover, if a community is ruled by rebel combatants (a rebel army or otherwise) that also does not count as a civilian rule of law. Thus, in order to broaden the definition and accurately capture the world we observe, we cannot assume that a community has some preexisting civilian rule of law, but rather, some local rule of law.
Furthermore, the word “temporary” is vital to the definition because martial law should last only as long as the emergency lasts. If these deviations from lawful, constitutional rule are permanent, and there is no intent to return to normalcy, then that situation represents a change in regime type rather than some triggering of an emergency response.
Finally, the first clause of the definition attaches an important and necessary condition: martial law must be explicitly declared and legally justified by the government. Emergency powers procedures are typically defined within the parameters of a state's constitution. Sometimes the constitution allows for martial law to expire within a certain time frame unless otherwise renewed by the executive or some legislative body at the higher levels of government. The declaration itself, especially including a legal justification, can be considered a way for a leader to cue both the public and fellow government officials of an intent to act in accordance with their constitution(s) while the emergency is ongoing and an intent to return to normalcy after. Without a declaration, such a replacement of local authority is not necessarily an emergency response but is arguably an attempt at brazen and unlawful repression. Some of the practices may involve purposeful repression, but the declaration itself is a signal that the government intends to act within legal parameters and uphold important norms. Indeed, rulers have incentives to adhere to legal constraints so as to maintain stability in the government (Weingast, 1997).
Moreover, the declaration serves a practical purpose, so that the executive can enter into the record when and where martial law is taking place. Martial law is mentioned in the analysis of judicial opinion by Ballantine (1912), where he cites Justice Holmes in the US Supreme Court case Moyer v. Peabody (1909). The Court includes in its opinion a quote of particular use: “Martial law exists wherever the military arm of the government is called into service to suppress disorder and restore public peace.” The resort to military aid necessarily means that the “rule of force under military methods is substituted” for civil authority “to whatever extent may be necessary in the discretion of the military commander,” and in respect to the commander's powers “there is no difference between a public war and a domestic insurrection.” (532)
To conclude this section and reiterate, martial law is an explicit and legally justified declaration by the central or regional government allowing the temporary substitution of local rule by the military. A military presence without such a declaration is not martial law, and a declaration that is not followed by a military presence is similarly not martial law.
Measurement
This section will review the definition of martial law and discuss ways to measure it. In this study, martial law is defined as the replacement of local authority with military authority following a formal declaration from the central government of the state. These declarations can come from a democratically elected executive, or from an authoritarian military junta, or any regime type in between. As stated, one of the central pieces of martial law is a declaration: an explicit statement from the government asserting control over a discrete area. Thus, an observable, measurable phenomenon is a declaration by the government that shows where and when martial law or a state of emergency would take place. Local or international news sources would be notified of the declaration, which would make the information public.
Therefore, the novel data presented in this article focus on de jure martial law, because attempting to find martial law events without declarations presents measurement problems. Measuring martial law by state behavior alone is quite difficult, if not impossible. Practices under martial law vary considerably, and there seems to be no base set of actions that qualify state behavior as martial law beyond military rule. Some examples of martial law in practice could include detention without cause, enforcing curfews, random (un)lawful search and seizure, or even arbitrary violence. Media blackouts may also occur. All or none of these manifestations of can happen under martial law. The practice and consequences of martial law vary widely from state to state, from executive to executive, so it is difficult to describe common practices beyond a generic description of military rule that may or may not result in reductions of personal liberties. While frustratingly vague at the outset, this also represents an opportunity to study martial law on a deeper level.
Often, martial law is directed at cities or districts, rather than on a countrywide basis as is typically thought. One of the emergencies that a state can use to justify martial law is rebellion. For example, as a response to the Kurdish conflict in Turkey, the Turkish government imposed a state of emergency in 13 provinces, to varying degrees, from 1987 to 2002. While the Turkish constitution drew a distinction between martial law and a state of emergency, the state of emergency allowed regional governors to extract and resettle entire communities in the regions under the order (Jongerden, 2007). In other words, it was inefficient to use military resources to cover the entire country of Turkey. Instead, repression was targeted at the regions where the conflict was occurring and where conditions were most unstable.
Creation of the dataset
This article offers a novel dataset documenting emergency powers events, including both states of emergency and martial law. To start, I use existing data that catalogs news articles and sorts them by event type. The Integrated Crisis Early Warning System (ICEWS) (Boschee et al., 2015) compiles hundreds of thousands of sources to build events datasets. One of the events that ICEWS catalogs is when a government declares a state of emergency or martial law (one event), and when it eases a state of emergency or martial law (another event). ICEWS does not differentiate between states of emergency and martial law.
After cleaning and sorting the ICEWS data, I found that there were some inconsistencies that had to be addressed before using the data in earnest. I trained and worked with a team of coders to triangulate the declarations by searching news databases, country by country and year by year, to find specific emergencies. We also undertook manual searches outside of the ICEWS events and found more instances of martial law and states of emergency. The final dataset therefore comprises manual coding of individual declarations sourced from news publications.
Because of this undertaking, we were able to make distinctions between the stated causes of the states of emergency or martial law and categorize them. The categories include violence/unrest, natural disaster, disease, humanitarian crisis, and other. Situations categorized as violence/unrest ranged from protest to outright violence. For example, in February 2022, Prime Minister Trudeau declared martial law in Ottawa in order to quell a COVID-19 related protest staged by commercial truck drivers (Carter, 2022). In January 2004, the government of Thailand declared martial law in three southern provinces as rebel violence escalated (Mintier, 2004). Humanitarian crises mostly comprise migrant and refugee crises, where governments used emergency powers in attempts to restore order at border crossings. Natural disasters and disease-related emergencies are relatively self-explanatory, and the final “other” category caught remaining unusual cases, including when Russia's government was forced to declare a state of emergency in the Siberian region because of wolf attacks (Oliphant, 2013).
The final empirical dataset moves beyond the ICEWS entries and includes data from 172 countries over 27 years, from 1995 to 2022. Over those years, 156 unique countries declared martial law or a state of emergency at least once, leading to a total of 2142 declarations. Of those declarations, 82 were explicitly martial law. Indeed, state of emergency declarations far exceed martial law declarations; states of emergency comprise 96% of the data while martial law comprises only 4%.
The dataset is disaggregated into months and includes the following variables: a count variable recording how many declarations happened during the month, how many emergencies were lifted, the areas mentioned in the declaration, the number of areas, the issue cited in the declaration, the declaration type, and whether the declaration was an extension of an existing declaration. For example, in December 2022, Jamaica extended a state of emergency once owing to violence and unrest in Clarendon, Saint Catherine, Westmoreland, Hanover, Kingston, Saint Andrew, Saint Ann, and Saint James (International Business Times Australia, 2022). Notably, if a state of emergency or martial law declaration applies to the entire country, then all of the country's regions are included in the variable recording the number of areas.
While the dataset does not identify cases in which martial law could have been declared but was not, constitutions and legal frameworks often outline the conditions under which such declarations are legal. These conditions typically include threats such as foreign invasion, insurrection/rebellion, or a breakdown of public order. These provisions outline situations in which martial law is in principle a viable option, even if no event is coded in the following dataset. This data does not create a catalog of non-events but instead provides users with a clearer baseline for interpreting the observed declarations. In this way, the dataset remains focused on actual instances of martial law and states of emergency while offering guidance for researchers interested in comparing declarations against their broader constitutional opportunity structures.
Emergency declarations, which include both states of emergency and martial law in response to violence/protest/unrest account for 32.6% of the data and natural disasters account for 39.6%. Declarations related to disease account for 20.4%, and finally, humanitarian crises and other unusual crises comprise 3.1 and 4.4% of the data, respectively. Importantly, however, of the total number of martial law declarations, nearly all of them were declared for reasons of violence and unrest, save one: in May 2000, Mongolia declared martial law in Dornogovi as a response to an outbreak of foot-and-mouth disease (Foot-and-Mouth Disease Kills 50 Animals in Mongolia, 2000). Additionally, 33 of these orders of martial law specifically covered the entire country, rather than were aimed at specific districts. That means that 40% of all martial law declarations were nationwide (Table 1).
Distribution of declarations by category.
Figure 1 shows a map displaying the relative frequency of declarations regardless of type. Figure 2, however, shows a map displaying the frequency only of declarations related to violence and unrest. Figure 3 shows a map of only where declarations of martial law occurred, derived from the larger dataset.
All declaration types.
Violence/unrest declarations.
Martial law only declarations.
Table 2 shows the top 10 countries that declared states of emergency or martial law for violence and unrest between 1995 and 2022.
Country counts of states of emergency and martial law declarations for violence.
Figure 4 shows the frequency of all declaration types over time. Notably, there is a large spike in 2020, owing to the COVID-19 pandemic. In 2020, 81% of declarations of emergency powers were related to disease. However, declarations remained relatively high afterward.
Frequency of declarations by year.
Figure 5 shows the frequency with which declarations cover areas of the country. Because emergency powers events do not always cover 100% of the districts or administrative regions within a country, we can visualize how often declarations cover certain percentages of the country's total area. There is a notable left-skew with a spike at 100% coverage. 2
Declaration coverage.
Figure 6 presents a subnational map of Colombia, as an example. It demonstrates the regional distribution of emergency power declarations from 1995 to 2022. The map highlights that the use of emergency powers is highly localized, with certain departments exhibiting many more declarations while others show fewer. This spatial variation underscores the importance of examining emergency powers at the subnational level, as national-level summaries alone would obscure these regional patterns. Notably, this map includes all types of declarations, such as natural disasters, violence and unrest, disease, etc.
Subnational variation of declarations in Colombia.
As with any data, limitations exist and should be addressed. Because I used news sources to find and record emergency powers events, the dataset is limited by exactly what information was reported. That means that for some declarations, there are not specific regions where the declaration was aimed. More importantly, news sources are far less likely to report when a state of emergency or martial law was lifted. The dataset does include a variable to count the number of times a declaration was lifted in a given month; however, it is probably not capturing the complete list of when emergencies ended. Researchers should therefore exercise caution when inferring how long a declaration was in place.
Comparison with other datasets
Table 3 below compares the scope, structure, and coverage of this dataset with other existing datasets on emergency powers. Unlike earlier efforts, this dataset offers a more granular view of emergency rule by recording declarations at the monthly level and identifying both the targeted region and the stated reason for the declaration. This design enables the analysis of more localized emergencies that would be obscured in country-year formats, as well as distinctions between emergencies enacted for public health, security, or natural disaster purposes. The data presented in this paper has more detail, which allows for more precise measurement of variation in emergency rule and subnational dynamics of emergency governance.
While prior datasets have made important contributions to the study of emergency powers, there are opportunities to improve temporal granularity, spatial detail, and country coverage. For example, data from Aydin and Gray (2023) records the start and end dates of emergencies but covers only 80 countries between 1918 and 2005, omitting more recent decades. Data used by Bjørnskov et al. (2022) include global coverage of up to 150 countries and includes comments on targeted region, and whether the emergency was constitutional, but reports only whether an emergency occurred in a given country-year and does not differentiate between a declared state of emergency and martial law. Hafner-Burton et al. (2011) is a widely used source that also records whether an emergency was caused by a natural disaster, but their data is derived from US State Department reports and contains a binary variable of whether a state of emergency was declared in a given country-year.
In contrast, the dataset in this project includes 172 countries from 1995 to 2022 and captures emergency declarations in a format that enables both cross-national comparison and fine-grained temporal analysis. This expanded coverage and detail offer a strong foundation for researchers examining how emergency rule plays out. Moreover, I include considerable data where there was not any sort of declaration, reducing the potential for observation bias; it systematically captures periods without such declarations to provide a clearer picture.
The combination of datasets used in this research substantially improves the ability to study emergency powers with precision, breadth, and historical context. Researchers can examine both contemporary trends and long-term patterns in emergency governance. The expanded country coverage, spanning 172 countries, and the inclusion of details such as the targeted region and stated reason for each declaration, offer a level of precision that complements coarser measures from prior datasets. A multi-source approach not only mitigates the limitations inherent in any single dataset but also enables new lines of inquiry into emergency rule including martial law, yielding a richer and more comprehensive understanding of how and why states invoke extraordinary powers.
Applications
To demonstrate how this data can be applied, I use it to explore whether rebel group activity increases the use of emergency powers. This question is motivated by the expectation that governments facing organized internal threats may invoke emergency provisions to consolidate authority or mobilize coercive capacity in response to instability. If emergency powers are employed as a means of managing or deterring insurgent violence, then I expect a positive relationship between the number of active rebel groups and the frequency of emergency declarations.
Indeed, these declarations are one tool that a government can use to respond to a security threat, such as a rebel group. Emergency rule is often framed as a tool for managing or containing conflict, granting executives greater authority to respond to threats. However, such powers can also escalate tensions by increasing repression or signaling regime instability. In some cases, emergency declarations may reshape rebel behavior, harden opposition, or alter the trajectory of ongoing violence Aydin and Gray (2023). A very preliminary regression model indicates that as the number of rebel groups fighting increases, the number of martial law/state of emergency declarations specifically related to violence is expected to increase. The dependent variable is a count of how many declarations of martial law and states of emergency occurred in a year, which ranges from 0 to 31. However, there is a high volume of zeroes, meaning that there are many years in which a country did not declare a state of emergency or martial law. To account for the overdispersion caused by high volume of zeroes, Table 3 shows a negative binomial model with two-way fixed effects and clustered standard errors at the country level. 3 The presence of rebel groups is therefore a strong predictor of emergency powers events in this initial, relatively simple model. Importantly, this regression focuses specifically on the conditions under which emergency rule is declared, rather than its downstream effects on conflict. Because the dataset captures the number of declarations rather than the number of days a declared emergency is active, this model does not capture the full dynamic of how a state responds to active rebel groups, as that falls outside the scope of these data (Table 4).
Comparison of emergency powers datasets.
Negative binomial with two-way fixed effects and clustered standard errors.
*** p < 0.001; **p < 0.01; * p < 0.05; † p < 0.001. Preliminary Model. 1995–2022. Country-year .
This dataset also opens new avenues for empirical research on emergency governance, civil–military relations, and state responses to crises. Researchers can apply the data to test competing theories about executive power, repression, democratic resilience, and the role of the military in domestic politics. Beyond theoretical contributions, the dataset facilitates comparative analysis, policy evaluation, and the integration of martial law into broader studies of states of exception, thereby addressing questions that have previously been limited by case-specific or anecdotal evidence.
More specifically, these data can help political scientists and policymakers understand critical political phenomena because it can be paired with other events data. In 2022, Cordell et al. published an updated dataset that used machine learning and text analysis to measure subnational repression, using allegations of human rights abuses at the subnational level for 196 countries (Cordell et al., 2022). Drawing on reports from Amnesty International, Human Rights Watch, and the US State Department, the authors compile hundreds of thousands of allegations of human rights abuses from 1996 through 2016. The martial law dataset can add to this existing universe of knowledge by adding explanatory power to these human rights abuse allegations. A comparison shows that declarations of martial law and states of emergency have a weak positive correlation of 0.15 with allegations of human rights abuses. A negative binomial regression with two-way fixed effects and clustered standard errors shows a positive relationship, although statistically insignificant. When looking deeper, Figure 7 shows an increase, and a cluster of human rights abuses as emergency declarations for violence and unrest move from 0 to 1. However, as declarations increase beyond 1, subnational repression appears to taper off. The initial declaration may be associated with a noticeable rise in abuses, but subsequent declarations may not be additive. This pattern points to a potentially nonlinear relationship that warrants further investigation. This plot is overlayed with the predicted counts of human rights abuses based on the model above (Table 5).
Abuses, declarations, and predicted counts of abuses.
Human rights abuses and violence emergency powers: two way fixed effects and clustered standard errors.
*** p < 0.001; ***p < 0.01; ** p < 0.05; † p < 0.001. TWFE and Clustered SE. 1996–2016. Country-year.
The emergency powers dataset introduced here makes best efforts to identify the subnational region in which the declarations occur, so there are opportunities to match these declarations with where the allegations relate to. The relationship between emergency rule and human rights violations remains an open question. While the two may certainly be connected, the precise mechanisms linking them are complex and contested. Rather than drawing strong conclusions here, I present this dataset as a resource that can help future research engage this debate more directly.
Data on declarations owing to violence or unrest can also be paired with data on one-sided government violence (Eck and Hultman, 2007; Pettersson et al., 2019). Doing so can give insight into the following questions: do emergency powers events inhibit or increase government violence in targeted areas? Are there more incidents of one-sided government violence in and around emergency powers zones, or does the declaration increase international scrutiny, which might decrease one-sided government violence?
Further, this dataset includes states of emergency after natural disasters, disease, and humanitarian crises that can be used to validate and compare with country level measurements for quality of governance. Does a better functioning government declare states of emergency more or less frequently after natural disasters? If a state of emergency for a natural disaster guarantees an influx of funding to the regional government, is there a proclivity to capture this funding for personal use, i.e. corruption? Are states of emergency or martial law used to garner support for candidates running for office, or worse yet, suppress voter turnout and guarantee an incumbent victory? Researchers can begin to address these questions with this dataset.
Conclusion
The “state of exception” is a critical political tool, one that can be used to consolidate power. Martial law is used with relative frequency, and yet its use remains poorly understood. As a policy, martial law does not remain in the past as some archaic, draconian measure of last resort. It is a viable option for governments to use in the face of an emergency, and clearly, governments use it relatively frequently. This article first offers an explicit definition for martial law: an explicit and legally justified declaration by the central or regional government allowing the temporary substitution of local rule by the military. Additionally, this article introduces a novel dataset documenting declarations of martial law and states of emergency. These declarations have clear and frequent use across the world. Notably, there is a marked spike in declarations in 2020 because of the COVID-19 pandemic, but there has also been a steady increase in declarations over time from 1995 through 2022.
This article makes a substantive contribution to the political science literature by creating opportunities for new insights into legalized state repression. From wolf attacks to civil war, from earthquakes to migration crises, governments have to respond to all kinds of emergencies. Both the definition offered in this paper and the data will allow researchers to better understand these events at the subnational level. Indeed, both political science researchers and policy researchers can make use of these data. Because the dataset makes best efforts to identify exactly where declarations were targeted, country experts can better understand the way some governments consistently deal with specific regions, and dive deeper into the conditions of human rights abuses. This relationship will require deep knowledge of cases, and this dataset offers a productive jumping off point. We can also look at whether there are neighborhood effects for emergency powers events; do neighboring countries use martial law on border regions when they are in conflict with one another?
Studying martial law and states of emergency will help scholars understand state responses to a wide variety of emergencies, including security threats as well as natural disasters and other emergencies. Furthermore, there are opportunities to use this data to evaluate individual country responses to emergencies, whether responses differ by regime, by leader, by military capability, or over time. We can question whether natural disasters help would-be autocrats seize power as some scholars have found (Bjørnskov et al., 2022). Additionally, these data will allow scholars to zoom into the subnational level to determine whether rural areas receive more or less attention when faced with an emergency. Finally, researchers can evaluate whether martial law, declared for violence or rebellion, provokes an international response from powerful countries. If so, there may be important network effects if grave human rights abuses under martial law or emergency rule disrupt reputations of powerful countries, and therefore, the alliance networks that stem from them.
Importantly, this data is limited by what information was reported and publicly available. However, the crux of the definition is that declarations must be made public anyway, as a signal to both opponents and allies to act within the letter of the law. There are always opportunities to update the dataset as more information becomes available.
In closing, martial law and states of emergency are events worth examining with a fine-toothed comb. Understanding preceding emergencies (or lack thereof) will give insight into the way leaders behave both in and out of times of crisis. In an increasingly volatile and conflictual world, examining the dynamics of emergency powers is essential for safeguarding democratic principles and ensuring accountable governance.
Footnotes
Consent to participate
Not applicable.
Consent for publication
Not applicable.
Data availability
Declaration of conflicting interest
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Ethical considerations
Ethical approval was not required for this manuscript.
Funding
The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the University of California Institute on Global Conflict and Cooperation (IGCC) Dissertation Fellowship. The views expressed in this manuscript are solely those of the author and do not necessarily reflect those of IGCC or its affiliates.