Executive Summary
Few conflicts ended in 2021, and many intensified. Overall levels of political violence remained steady, while its lethality surged to new heights. Civilians bore the brunt, with increases in both violence targeting civilians and civilian fatalities. Where conflict declined last year, fragile ceasefires often obscured the growing risk of escalation — now seen in Yemen and Ukraine. Where long wars ended, such as in Afghanistan, new risks have emerged for civilians.
Against this backdrop, social unrest continued to build during the second year of the COVID-19 pandemic, with a rise in demonstrations all around the world from rallies against public health restrictions to mass pro-democracy protest movements. Authorities responded violently in places like Myanmar and Colombia, leading to a spike in demonstrator fatalities.
As the world struggles to come to grips with a third year of the pandemic — as well as the potentially devastating consequences of Russia’s invasion of Ukraine — global insecurity is likely to only deepen and evolve throughout 2022 (for more, see ACLED’s special report on 10 additional conflicts worry about).
Over the course of 2021 and early 2022, ACLED expanded to Canada, Oceania, Antarctica, and all remaining small states and territories, bringing the dataset to full global coverage. Our 2021 annual report reviews the past 12 months of data on political violence and demonstration activity around the world. Key findings are outlined below.
Key findings
Political violence remained at similar levels relative to 2020. Political violence decreased by less than 2% — or 1,449 events — in 2021. In total, ACLED records 94,833 political violence events in 2020 and 93,384 in 2021. While regions like Central Asia and the Caucasus, Europe, and South Asia experienced significant decreases in violence, Southeast Asia saw an increase of 190%, driven by the military coup and ensuing crisis in Myanmar.
Reported fatalities rose significantly. Fatalities from political violence increased by 12% — or 15,740 deaths — compared to 2020. In total, ACLED records 131,370 fatalities in 2020 and 147,110 in 2021. Political violence was deadlier in 2021 in every region except for Europe and the Middle East. Fatalities in Central Asia and the Caucasus increased by more than 11%, despite a decrease in political violence events.
Civilian targeting and civilian fatalities increased. In 2021, civilians were more frequently targeted by political violence than in 2020, with far deadlier consequences. ACLED records a 12% increase in civilian targeting overall last year, with 33,331 events reported in 2020 compared to 37,185 in 2021. Civilian fatalities too increased by 8%, with 35,889 reported fatalities in 2020 compared to 38,658 in 2021. The greatest increases in civilian targeting were recorded in Myanmar, Palestine, Colombia, Nigeria, and Burkina Faso.
Anonymous armed groups, as well as state forces, remained the most active perpetrators of violence targeting civilians. For the second consecutive year, anonymous or unidentified armed, organized groups top the list of most active perpetrators of civilian targeting, responsible for the largest proportion at 45% of all events reported in 2021. Of identified actors, state forces continued to pose the greatest threat to civilians last year, responsible for 16% of civilian targeting and 14% of civilian fatalities.
Despite the continued rise of violent non-state actors, including rebel groups and political militias, state forces remained the most active conflict agents. While overall state engagement in political violence fell substantially — by 18% or 10,195 events — for the second year, state forces remained the dominant conflict agents globally. State forces engaged in 46% of all political violence events in 2021.
Ceasefires obscured persistent fragility in 2021. The greatest decreases in political violence took place in contexts where ceasefires and power-sharing agreements were reached in 2020 and early 2021. These include: Azerbaijan and Armenia over the breakaway Artsakh Republic1; Ukraine and Russian-aligned separatists in Donetsk and Luhansk; Pakistan and India along the disputed Line of Control (LoC); and rival governments in Libya. Similarly, in Yemen, a decrease in violence in 2021 was driven by the formation of a power-sharing cabinet between the Southern Transitional Council (STC) and the Hadi government in the southern governorates in late 2020, alongside international engagement in peace negotiations on the Houthi conflict. Despite the declines in violence, each context remains deeply fragile and susceptible to further outbreaks of violence. In both Yemen and Ukraine, this fragility was readily apparent as fighting continued throughout 2021 and ultimately escalated in early 2022.
Demonstration activity continued to increase in the second year of the COVID-19 pandemic. Demonstrations rose by 9% — or 13,104 events — in 2021, compared with 2020. While this growth was driven by a range of movements, mass anti-government demonstrations were particularly prominent in those countries that registered the greatest increases in demonstration activity. These included mass demonstrations against health, education, social, and economic policies in Colombia; anti-coup demonstrations in Myanmar; water and wage-related demonstrations in Iran; and, demonstration movements against COVID-19 restrictions in France and Italy.
Glossary
In this report, the term political violence refers to all events coded with event type Battles, Explosions/Remote violence, and Violence against civilians, as well as all events coded with sub-event type Mob violence under the Riots event type. The latter is included given that this violence, while spontaneous rather than organized, is often similar in nature to violence involving communal groups such as local security providers. Mob violence is more similar to other forms of political violence than it is to violent demonstrations (described below). Including Mob violence within political violence also has the benefit of allowing for a better understanding of the spectrum of political violence and how it may manifest differently across different spaces.
The complement to political violence in the ACLED dataset is the term demonstrations, which is used in this report to describe all events coded with event type Protests, as well as all events coded with sub-event type Violent demonstration under the Riots event type. Expanding on the point raised above, Mob violence is not grouped with demonstrations here because it looks less like other forms of demonstrations associated with mass social movements, and more like political violence. The events included under demonstrations here are what users may typically associate with social movements — in which groups of demonstrators advocate for a certain policy or belief. These demonstrations may be peaceful or violent.2
The term disorder is then used in this report to refer to all political violence and demonstrations. This effectively includes all events in the ACLED dataset, minus Strategic developments — which should not be visualized alongside other, systematically coded ACLED event types due to their more subjective nature. For more on Strategic developments and how to use this event type in analysis see this primer.
Following from these, armed organized violence is a subset of political violence, made up of all events coded with event type Battles, Explosions/ Remote violence, and Violence against civilians. The difference between political violence and armed organized violence is that Mob violence is included in the former; it is not included in the latter as this violence is spontaneous in nature (not organized) and often does not include armed individuals.
Both the term conflict and war refer to campaigns of events, rather than specific event types. The ‘war in Yemen’, for example, may include a variety of types of events — Battles, Explosions/Remote violence, and Violence against civilians. These categorizations will hence be denoted on the basis of the actors involved, the location of events (at the country or subnational level), and/or time periods. The distinction between the two terms (conflict and war) is one of scale: the latter is a more intense form of the former. For more on this categorization, see the latter portion of this primer.
Civilian targeting refers to all violence which targets unarmed individuals. It is important to remember that events coded with event type Violence against civilians are only one subset of this violence. Civilians can also be targeted in events coded with event type Explosions/Remote violence and Riots, as can Protesters (i.e. unarmed demonstrators). Protesters can also be targeted through lethal forms of violence in events coded with sub-event type Excessive force against protesters coded under event type Protests.
While ACLED records fatalities, it is important to remember that the set includes reported fatalities. Fatality numbers are frequently the most biased and poorly reported component of conflict data. The totals are often debated and can vary widely. Conflict actors may overstate or under-report fatalities to appear strong to the opposition or to minimize international backlash against the state involved. Fatality counts are also limited by the challenges of collecting exact data mid-conflict. While ACLED codes the most conservative reports of fatality counts to minimize over-counting, this does not account for biases that exist around fatality counts at-large (for more, see: Washington Post, 2 October 2017). While fatality estimates are a telling indicator of how conflict intensity and lethality shift over time, they are generally less reliable than other metrics coded by ACLED, due in part to the highly politicized and varying fatality information reported by different sources. Such a metric is therefore largely used as a supplement to other modes of analysis, and is why it is treated as a measure of reported fatalities, rather than a concrete number. For ease of readability, this report often refers solely to fatalities, but please note that these are references to reported fatalities specifically. For more on ACLED’s methodology around coding fatalities, see this primer.
Footnotes
- 1
The disputed territory of Nagorno-Karabakh is internationally recognized as part of Azerbaijan. ACLED refers to the de facto state and its institutions as Artsakh — the name by which the de facto territory refers to itself. For more on methodology and coding decisions around de facto states, see this methodology primer.
- 2
While ACLED collects information on demonstrations, it is important to remember that these are demonstration events. ACLED is an event-based dataset, and therefore only records demonstration events; the number of ‘demonstration events’ recorded by ACLED may differ from the number of ‘demonstrations’ recorded via other methodologies. The number of demonstrations is reliant largely on reporting and the terminology used in doing so. For example, five separate demonstrations happening in Algiers around a single topic within a few blocks of each other may be reported on in a newspaper as “demonstrations happened in Algiers” or “five demonstrations happened in Algiers.” Both are correct in their terminology, but if they are coded differently as a result (1 vs. 5), this would introduce a bias. ACLED codes an event based on an engagement in a specific location (e.g. Algiers) on a specific day in order to avoid such biases. For ease of readability, these events are often referred to solely as demonstrations in this report.
