On June 1, the U.S. Defense Department (DoD) submitted to Congress its report on the more than 500 civilians it estimates were killed and injured in U.S. military operations in 2017—as required by Section 1057 of the National Defense Authorization Act (NDAA) of 2018. Despite several limitations and gaps in civilian casualty reporting (for example, it doesn’t cover strikes by the CIA or other agencies), the Pentagon’s report is in some respects a step forward from prior U.S. government disclosures on civilian casualties, for example by providing country-specific information and data on injuries, and is an important sign of the Pentagon’s commitment to release such data on an annual basis. As two of the authors explained in a report released last year, the release of more detailed information on civilian casualties helps the public to meaningfully assess the true costs of U.S. operations abroad and provides some basic recognition of the harm suffered by those injured and families of those killed.

Last month, we proposed a framework for Congress to evaluate DoD’s reporting on civilian casualties, and its tracking and protection methods and processes. We use that framework to provide some analysis of the report below, alongside an overall analysis of the significance and limitations of the report itself.

We found that future DoD reports should include more detail on specific incidents, including discrepancies between independent reporting on civilian casualties and DoD estimates, as well as inclusion of “combatant” deaths and details of investigations conducted and the outcomes of those investigations.

At the same time, the report needs to be situated in the context of broader concerns about civilian casualties. In the wake of well-documented and enduring concerns about increased civilian casualties, U.S. conduct in specific incidents, backsliding on transparency, and criticism of U.S. military investigations of civilian harm, a broader reckoning of the way the U.S. government tracks, investigates, and responds to civilian harm is needed. We identify some of the key steps that need to be taken, including improving the way the Pentagon counts casualties by lowering the requirements for a civilian casualty to be deemed “credible,” conducting site visits and witness interviews where feasible and explaining when it is not possible to do so, holding regular meetings with NGOs to discuss specific cases and civilian casualty methodology, and publishing reports on civilian casualties by all government actors, not just the military.

What does the report include?

1) The report references and endorses a key 2016 Executive Order on civilian casualties. The report specifically references and incorporates in large part an important 2016 Executive Order as “catalogu[ing] the best practices DoD has implemented to protect civilians” and “direct[ing] that those measures be sustained in present and future operations.” The Executive Order has its limitations, but was an important and positive step towards more transparent and accountable U.S. policy and practice on the use of force. The incorporation of the Executive Order into this report is an important sign that DoD considers the Executive Order relevant and applicable to its overseas operations in the wake of reports that the White House is considering “modifying” or “rescinding” it. 

2) Aggregated numbers of civilians killed and injured. The report provides an aggregated figure of civilian casualties that DoD assessed to be “credible.” Unlike the summary reports on civilian casualties released by the Obama administration in 2016 and 2017, the DoD report includes an estimate not only of the number of civilians killed in overseas U.S military operations, but also of those injured. The report estimates that there were 499 cases of civilian deaths and 169 injured civilians as a result of all U.S. military operations in 2017, and states that more than 450 reports from 2017 continue to be assessed—indicating that the civilian casualty estimate may, in fact, be higher. The figures include civilian casualty estimates for operations in Afghanistan, Iraq, Libya, Syria, Somalia, and Yemen.

3) Country-specific data. Notably, the report provides civilian casualty data in relation to specific countries in which the U.S. military is operating. This is an advance on previous reporting on civilian casualties, which provided just aggregate numbers for a year without reference to civilian harm in specific operational theaters. DoD acknowledges civilian casualties in Afghanistan, Iraq, Syria, and Yemen, but reports that there were no credible civilian casualties in Somalia or Libya from 2017 (although it acknowledges that one incident in Somalia remains under review).

4) A general overview of the credibility assessment and investigations process. Section II of the report provides an overview of how DoD reviews and investigates possible cases of civilian casualties. It notes that specific processes to review or investigate claims of civilian casualties depend on a range of factors that are specific to the command or operation, including host nation requests, mission objectives, and available resources or access to the ground where operations take place. The report then summarizes the process by which the military assesses the credibility of civilian casualty reports. As in past instances of public reporting on civilian casualties by the U.S. government, DoD explains that a claim of civilian casualties is examined against information “readily available” to the command.

As with previous U.S. government disclosures, the DoD report provides a general list of the external and internal sources it consults when reviewing possible civilian casualty incidents. These sources may include NGO reports, news and social media, as well as video surveillance and other data from intelligence, surveillance, and reconnaissance assets. The DoD report, however, goes further by stating that the Pentagon reviews witness observations, including that of partnered forces, although it does not specifically explain if this includes carrying out interviews with witnesses and victims, or site visits. The lack of witness interviews and site visits has been a consistent criticism of NGOs.

The report is notable for setting out explicitly that the U.S. military applies a “more likely than not” standard to determine whether civilian casualties are “credible.” This was recently acknowledged in a U.S. Central Command email to journalist Azmat Khan, but is perhaps the first time this has appeared in a public document of this nature. DoD also provide a specific example of a discrepancy between its own reporting and that of the United Nations Assistance Mission in Afghanistan. As noted below, more examples of this kind would be welcome.

5) An overview of civilian harm mitigation steps. The report reaffirms the “best practices” to minimize civilian harm as reflected in the 2016 Executive Order. Among other steps, the report outlines efforts to protect civilians during U.S. military operations in 2017, including by “characterizing the operating environment”; crafting the operational design to avoid civilians during ground operations; mitigating the effects of weapons on civilians and civilian objects; and optimizing targeting processes.

What was missing and what could be done to improve the Pentagon’s civilian casualties report in the future?

The report is a significant step forward towards transparency but falls short in several key respects. To build on the transparency in this report and other forms of public reporting, DoD should voluntarily provide a more granular level of detail in future iterations of the report, including: 

1) Disaggregated data. As noted earlier, the DoD report provides a total number of those it assessed were injured and killed across all U.S. theaters and does not break those figures down by month, country, location, age, or sex.  While we are aware that DoD provided disaggregated casualty estimates in its classified annex to Congress, the reasons for doing so remain unknown, and both the Pentagon and Congress should work towards declassifying this annex.

The provision of disaggregated data—in a centralized report, and in relation to each command or theater—would enable policymakers, NGOs, and other stakeholders to better understand where and why U.S. government figures are at odds with independent estimates. This is a recurring issue: For example, in 2017, independent monitors and organizations recorded estimates of civilian casualties as a result of U.S. lethal operations in Iraq and Syria alone that are far higher than the total number of deaths and injuries the Pentagon reported across all its operations. In Somalia, the DoD report says that there were no civilian casualties (with one case pending) despite reporting suggesting that the U.S. military was involved in a number of civilian casualties during last year.

2) More specific information on discrepancies with NGO figures. More information on specific cases and discrepancies would also help in understanding the differences between official U.S. military civilian casualty figures and independent estimates. As noted above, the Pentagon report is helpful in providing an example of how the U.S. military addressed a discrepancy in a case monitored by the United Nations in Afghanistan. More examples of this, particularly in relation to cases where there is significant independent documentation and reporting on alleged civilian casualties, would allow for a more nuanced discussion between civil society and the military on specific incidents, demonstrate how the military assesses the credibility of civilian casualty allegations in practice, and what measures it took to investigate and verify when there were strong allegations of civilian casualties. These should include whether or not site visits or witness interviews were undertaken, and, if not, explanations for why this was not possible.

3) Inclusion of “combatant” deaths and how it determines and defines “combatant” status. The report excludes the number of combatant deaths, noting that DoD has decided not to count the number of enemy combatants killed or wounded during operations by alluding to the practice of counting bodies during the Vietnam War. We agree that using a “body count” as a measure of operational success is misguided. However, the purpose of doing so here would not be to define a mission’s operational success, but rather to help establish how many civilians are being harmed in U.S. operations. The way that the Pentagon defines and determines the “civilian,” “non-combatant,” and “combatant” status categories is subject to an ongoing debate. By including the DoD’s definition of “combatant,” more information about how it determines “combatant” status, and figures for “combatant” deaths (especially if disaggregated, see below), would allow the public to understand how the term “combatant” is defined and understood by DoD, how DoD differentiates such a category from those killed whose status is undetermined or unknown, and reveal more about the discrepancies between DoD and NGO civilian casualty figures.

4) Include a category of cases where there is “insufficient information to make a determination.” As noted above, the report concedes that many allegations of civilian casualties are not deemed credible for lack of information, or while pending review. DoD could easily bring its own estimates closer to public numbers by defining the number of cases for which it simply has insufficient information one way to a category of “unknown,” rather than “non-credible.”

5) Detailed information on how specific investigations are conducted, the outcome of those investigations, and any lessons learned in the process. As we outlined in our previous post, DoD should release information on the actual investigations conducted, including the results and outcomes of such investigations on an ongoing basis and collated annually in the NDAA report, but the report does not provide this detail. Greater clarity from DoD on the lessons learned from specific operations, the constraints or factors that influence the conduct of investigations, and how mistakes can take place in spite of significant effort to avoid them matters to the families of those killed and survivors of U.S. strikes and bolsters DoD’s credibility, public legitimacy, and is demonstrative of its commitment to promote the rule of law and minimize civilian harm.

6) Officially publish the report on the Department of Defense website. Sending the report to specific NGOs was an important step forward in transparency, but it should not be up to those NGOs to disseminate further. The report should be published officially on the DoD website as a sign of its institutional commitment to transparency.

Broader concerns about civilian casualties

The DoD report provides an opportunity for the Pentagon to review how it may improve its methods to investigate and report claims of civilian casualties, even if in and of itself it cannot address all the concerns that have been raised around this issue. We offer a selection of recommendations here that the Pentagon and the U.S. government can take that would help improve transparency and the tracking and mitigation of civilian harm in U.S. operations.

  • Lower the threshold for assessing a civilian casualty to be “credible.” The DoD report explains that “credible” means “it is assessed that it is more likely than not that the report regarding civilian casualties is correct.”  In other words, it meets the “reasonably suspected” test.” As Ryan Goodman has noted, a key systemic flaw in U.S. military assessments of civilian casualties rests on the Pentagon’s use of an overly high evidentiary standard for the initial assessment of civilian casualties. The standard is too high because it is applied to the mere tracking and counting of civilian casualties: Credibility assessments are not the same as a finding of legal fault and do not result on their own in any administrative or criminal sanction. They are, however, important for acknowledging the civilian harm caused (even if lawful), so that lessons can be learned, and as a possible precursor to more serious accountability mechanisms where such action is warranted. It is notable that the DoD report itself implicitly acknowledges the problem with the evidentiary standard, when it states that “in some cases, DoD has not been able to assess a report as credible because insufficient information has been provided.” The problem here is with the standard that is applied before an investigation is conducted which at present prevents a comprehensive investigation into serious allegations of human rights and humanitarian law violations. Investigations should not be curtailed merely because the U.S. military has “insufficient information.” In fact, those cases may precisely be the ones requiring further and more thorough investigation. As such, the current standard should be re-evaluated and lowered by the Pentagon as a matter of official policy.
  • Where feasible, the U.S. government should conduct site visits or speak to witnesses where there are credible allegations of civilian casualties and, where not possible, provide an explanation for why it was not able to do so. As stated above, the U.S. government should seek out evidence from a wider range of sources by conducting site visits and interviews with victims and witnesses on the ground. The failure to conduct site visits and interviews has been an oft-repeated criticism of NGOs. As highlighted by a group of experts on Just Security last year, speaking to witnesses and victims provides investigators with a key source of information that can “provide important contextual details, resolve inconsistencies, and can shed light on people’s identities, motives, and intentions.”
  • The U.S. government should hold regular meetings with groups of NGOs and civil society, in theater and in Washington, to discuss civilian casualties. Conversations with groups who carry out detailed on the ground investigations and speak to victims and witnesses, may enable the military to improve their investigative practices in specific cases. Doing so may also help dispel critiques that U.S. military investigations and processes are secretive and incomplete, can shed greater light on the discrepancies between U.S. government and NGO reports of civilian casualties, and will allow for more productive dialogue that can help enhance the military’s ability to protect civilians. In some cases, these details may be better clarified in theater-specific contexts through other forms of civil-military engagement or public reporting. The model for this exists: It already happens with the United Nations in Afghanistan, and this could be expanded to include broader consultations in different countries.
  • The White House must report publicly on civilian casualties in all operations. The June 1 report covers only military operations, even though other government agencies are involved in carrying out lethal strikes. Accordingly, this report is not a substitute for broader reporting by the White House on civilian casualties in all of its operations, and it remains a concern that the Trump administration has not complied with its own separate civilian casualty reporting requirement under the Executive Order.

We acknowledge that all the limitations and gaps in civilian casualty tracking cannot be addressed through an annual DoD report to Congress alone. However, the report serves as a useful starting point for a constructive discussion about how to better acknowledge, investigate, and mitigate civilian harm in U.S. military operations.

 Photo by Chris McGrath/Getty Images