By Mollie Zapata, Simon-Skjodt Center for the Prevention of Genocide
The Early Warning Project asked, “Between 1 April 2017 and 31 March 2018, will an armed group from [Country] engage in a campaign that systematically kills 1,000 or more civilians in [Country]?” for 17 countries we determined to be at high risk for onset of mass killing.
To resolve our questions, we draw on multiple sources, depending on what information is available and most relevant for the country in question. We first look at any publicly available datasets, including global datasets like the Armed Conflict Location Dataset (ACLED), and country-specific datasets like Iraq Body Count and the CFR Nigeria Conflict Tracker. Because these datasets rely on media sources for information about specific incidents, we assume they tend to undercount fatalities when and where access is significantly curtailed and/or media are limited or unfree. These datasets also don’t cover every country. We therefore also look at the reporting and analysis of groups, including the United Nations, US government, Human Rights Watch, International Crisis Group, and others, which sometimes include estimated fatalities from an episode of violence or series of attacks, not solely individual incidents. When there is a “close call” or the fatality totals vary widely between sources, we may also call upon experts to assist in our decisions.
As noted in the question description, armed groups include both state forces and non-state groups like rebel armies and militias. Campaigns that systematically kill civilians include, but are not limited to, policies which intentionally kill civilians en masse (e.g., military strategies that intentionally target civilians, mass execution of a specific group) and policies that knowingly result in widespread death (e.g., mass starvation, confiscation of health care supplies, forced relocation). In general, unrelated executions of individuals or the accidental killing of civilians in war (“collateral damage”) will not be considered a campaign to systematically kill civilians. If an armed group is engaged in multiple campaigns that systematically kill civilians (e.g., in different geographic areas, or targeted against separate civilian groups) those fatalities will be counted separately and the question will only resolve as yes if 1,000 civilian fatalities occur in one or more campaigns. See EWP for examples. See here for GJ’s FAQ on forecasting questions like this.
Explanations for Close Calls that Resolved as “Yes”
Iraq — Yes
In the battle for Mosul (October 2016 – July 2017) ISIS engaged in a campaign that systematically killed more than 1,000 civilians, suspected of being opposed to ISIS. The Associated Press (AP), using data from Amnesty International, Iraq Body Count, and a United Nations report, claims that between 9,000 and 11,000 people were killed during the battle. They estimated a third died as a result of intentional targeting by ISIS. Though the battle began in October 2016, from the time we begin counting in April 2017, AP reports over 2000 civilian fatalities as a result of ISIS actions in Mosul. As in Yemen, fatalities caused by a foreign-led force (in this case the US-led coalition) do not count toward the total, as we require the perpetrator to be a group from the country in question. As an additional source, ACLED counts 2459 civilian fatalities as a result of ISIS violence in the 12 month period.
Myanmar/Burma — Yes
Based on a survey of refugees and established epidemiological methods, Médecins Sans Frontières (MSF) estimates that 6,700 Rohingya civilians were killed in Burma by military and other security forces in a government campaign—“in the most conservative estimations”—due to the violence that began in August 2017. This contrasts with much lower estimates from sources based on news reports of specific incidents, such as the Political Instability Task Force (PITF) Worldwide Atrocities Dataset. This discrepancy is unsurprising, since the government has systematically restricted access to the region. Given the large numbers of refugees (650k) and IDPs (150k) as well as the satellite imagery evidence of the destruction of hundreds of Rohingya villages and UN estimates around 1,000 deaths in September of 2017, we deem the higher estimates to be credible.
South Sudan — Yes
Despite very limited news reporting from South Sudan, ACLED counts 1176 civilian deaths in the 12 months in question. Notably, a February 2018 report from a UN-sponsored commission of inquiry stated, “The brutality of attacks against civilians has not been limited to direct attacks on their lives but importantly has also included the systematic looting and burning of villages, destroying people’s sense of security and ability to support and care for themselves. As a result, millions of citizens have been displaced, and thousands are sheltering in the bush, resulting in untold deaths from starvation, thirst, exposure, and lack of access to medical care. These deaths are a direct and foreseeable result of the conflict, and no less part of the war’s casualties than those shot, beheaded, burned in their tukuls, or strung up from a tree.” Similar to Myanmar, the number of displaced and the minimal reporting coming from the areas most affected by violence suggest that fatality levels are much higher than reported in international data aggregation, leading us to conclude that at least 1000 civilians from various ethnic groups but all categorized as “anti-government” were targeted by government soldiers in South Sudan.
Philippines — Yes
In its 2018 annual report, HRW claims that “more than 12,000 suspected drug users and dealers, mostly from poor families in urban centers across the country, are estimated to have died” from July 1, 2016 through September 26, 2017. Previously, in March 2017 (before our question period began), HRW published a report stating that 7,000 people, mostly poor drug users, had been killed by police and vigilantes in the war on drugs. Thus, HRW estimates that approximately 5,000 have died in the time period in question. Although it is unknown what proportion of these fatalities was the result of systematic government action, the HRW numbers being so high above the 1,000-death threshold and evidence of a continuing systematic government campaign against drug dealers and users leads us to determine that this questions should be resolved as ‘yes.’ (In our annual 2017 report we determined that there was an onset of mass killing in the Philippines in 2016, see more in our 2016 State of the World report.)
Explanations for Close Calls that Resolved as “No”
Afghanistan — No
Determining total civilian fatalities in Afghanistan is particularly challenging due to the nature of the conflict and reporting on it. According to ACLED, from April 2017-March 2018, 612 civilians were killed by the Taliban, which is the largest perpetrator group in terms of civilian deaths in the country. UN estimates suggest that in total 1443 civilians were killed by the Taliban in 2017 and that the level of violence was similar through the first quarter of 2018. Though this number is above the 1000 threshold, the UN reports fatalities due to ground engagements, suicide attacks, remote IEDs, suicide IEDs, targeted killings, and abductions. We require fatalities to be part of a “systematic campaign” against civilians, and there is insufficient evidence that more than 1000 civilian fatalities were the result of intentional civilian targeting. For example, civilian fatalities in “ground engagements” might be incidental to attacks on military targets, and remote IEDs are usually placed on roads frequented by military convoys (implying that civilians may not have been intentionally targeted).
Democratic Republic of the Congo — No
While multiple sources indicate that total civilian fatalities through the Democratic Republic of the Congo is high (e.g., ACLED reports 1384 civilian fatalities in the time period in question), there is a large number and diversity of perpetrators.The Kivu Security Tracker lists 126 armed groups currently operational in North and South Kivu alone. According to the ACLED data, the perpetrator group responsible for the most civilian fatalities was the Bana Mura, killing 261. Thus we determine that no one perpetrator group is responsible for more than 1,000 civilian fatalities in the time period in question.
Nigeria — No
The Council on Foreign Relations’ Nigeria Security Tracker records 982 total deaths (total, not only civilians) by Boko Haram in the time period in question. ACLED recorded that Boko Haram killed 837 civilians. Similarly, fatalities perpetrated by Nigerian security forces did not reach the 1,000-civilian fatality threshold, according to CFR’s tracker. We also evaluated violence perpetrated by sectarian actors: CFR’s tracker totals fatalities by “sectarian actors” at 1236; however, the category includes multiple actors, all of which are not working together in a coordinated campaign. There is no evidence that a single group is engage in a campaign that resulted in 1000 civilian fatalities from a particular target group. Thus we conclude that Nigeria did not reach the 1,000-civilian fatality threshold from April 2017-March 2018.
Sudan — No
According to ACLED, civilian fatalities as a result of government or militia violence were relatively low in the 12 months in question (43 total), while according to PITF they totalled to 110. Using either source, there is insufficient evidence that Sudan experienced a mass killing.
Yemen — No
Based on available information, civilian fatalities caused by Houthi groups or ISIS do not appear to reach the 1,000-civilian fatality threshold in the 12 months in question. ACLED lists 45 civilian fatalities perpetrated by nonstate groups in the time period, in comparison to 110 state-perpetrated civilian fatalities. Though these numbers are undoubtedly lower than the true numbers, there is no strong indication that a group in Yemen has intentionally killed more than 1,000 civilians in Yemen. (See more on how we count fatalities in Yemen here.) Finally, evidence suggests that fatalities due to blockades and supply shortages are mostly due to the Saudi-led blockade. These deaths are excluded because Saudi Arabia is a foreign actor.
Burundi — No
Central African Republic — No
Mali — No
Pakistan — No
Turkey — No
Ethiopia — No
Bangladesh — No
Zimbabwe — No