The Pentagon’s latest annual report on civilian casualties, released this morning, lists shockingly low numbers of reported casualties that the Department of Defense has assessed as “credible,” reflecting a need for better processes for determining how many civilians the United States is killing. While the report contains some improvements in reporting standards and parameters, it illustrates the serious gaps that remain to be address by both the department and Congress.

The department issued the report to comply with provisions in the last two defense authorization bills. This article provides readers with a refresher on the contents of the original provision in the Fiscal Year 2018 National Defense Authorization Act (NDAA), the new requirements added in the 2019 bill, and how DoD responded to them in this latest report. (Readers can also look back at an analysis last June of the 2017 report.)

The May 2019 report is the second of its kind and the first report after Congress passed more robust reporting requirements in the FY 2019 NDAA. The latest report, due to the armed services committees yesterday and first reported by Missy Ryan in the Washington Post, was posted on Defense.gov this morning. It builds on a record of incremental progress toward greater transparency from the Pentagon on its record — and its methods — with respect to preventing, tracking, and responding to civilian casualties.

Importantly, the latest report helps surface some of the toughest remaining issues worthy of attention from DoD, such as the process that leads to such ostensibly low casualty figures. As it works to improve the accuracy of identifying civilians, both for preventing civilian casualties and for tracking those that occur, the Pentagon should work closely with outside experts and organizations that advocate for greater protections for civilians in war and against overly expansive definitions of who is targetable.

What the NDAA Requires 

The reporting requirement, which originated in Section 1057 of the 2018 NDAA, called on DoD to provide a list of the operations that were confirmed, or reasonably suspected, to have resulted in civilian casualties. The original provision also required the date and location of each operation, and whether or not the operations took place inside or outside a declared theater of active armed conflict.

Last year’s NDAA (Section 1062) amended the requirement in a few key ways designed to increase the level of detail into certain aspects of DoD’s analysis and to ensure that the report, including key information, would be made public at the same time that Congress received it.

The most notable revisions in the 2019 amendment (Sec. 1062) include:

  • The addition of “each specific strike, mission, raid, or incident” to the general requirement to provide a list of each operation that was reasonably suspected or confirmed to have resulted in civilian casualties.
  • The addition of a requirement to differentiate between those harmed and those injured in the department’s tallies.
  • The addition of a requirement to describe the process by which the department makes ex gratia payments (in addition to the description of how DoD investigates allegations).
  • A requirement to make public the report submitted at the same time as it is provided to Congress.

This Year’s Report 

In some ways, this year’s report maintains a degree of consistency with last year’s – much of the prefatory language remained the same or similar, while the addition of a table of casualties in each section marks a significant difference and an improvement over last year’s report.

Total Figures and Estimates Broken Down by Location:

As with last year’s report, DoD provides an aggregate figure of the overall estimated casualties worldwide. (Last year’s report provided an aggregate figure of 499 killed and 169 injured.) This year, DoD claims that 120 civilians were killed and approximately 65 civilians injured during 2018 as a result of U.S. military operations in Iraq, Syria, Afghanistan, and Somalia. The report states that the department found no credible reports of civilian casualties from U.S. military operations in Yemen or Libya (although one report in Libya from 2017 remains under investigation.) By contrast, tracking organizations (Airwars, UNAMA, Bureau of Investigative Journalism, and New America) assess with confidence that at least 1,224 civilians were killed (including 8 in Yemen and 3 in Libya) – a figure that exceeds the DoD estimate by a factor of 10.*

But unlike last year’s report, it also provides a breakdown of total figures by country (or theater):

  • In Iraq and Syria (Operation Inherent Resolve), DoD assesses that 42 civilians were killed and approximately 7 civilians injured as a result of U.S. military actions while 28 reports had not yet been assessed. The report also includes an assessment that approximately 793 civilians were killed and approximately 206 civilians were injured as result of U.S. military actions in 2017, while 69 reports were still under evaluation. (Airwars assesses 805 civilians killed during the same time period.)
  • In Afghanistan (Operation Freedom’s Sentinel and NATO-led Inherent Resolve), DoD assesses that its actions resulted in 76 civilians killed and approximately 58 civilians injured. (UNAMA assesses that 406 civilians were killed during the same time period.)
  • DoD assesses two civilians were killed in Somalia (operations against Al Shabaab) as a result of U.S. military actions. (The Bureau of Investigative Journalism also assesses that at least two were killed in Somalia during this time, but Amnesty International has recently contested the accuracy of figures). 

Figures broken out by location, type of operation, and casualty:

Unlike last year, and true to the reporting requirement, the report includes a breakout of operations in each theater in which they conducted lethal operations, along with whether the operation was a ground or air engagement, and whether the casualty was an injury or death. While it is possible that each theater may have independently reported this information previously, this year’s report is an improvement over last year’s, which provided aggregate figures without a more structured breakdown. 

Greater detail on the credibility assessment process: 

This year’s report contains more details about the process by which a command assesses the credibility of a report of civilian casualties, which is the threshold the Pentagon uses for tallying the number of confirmed or “reasonably suspected” casualties:

A report may be found to be “not credible,” if, for example, (1) there was no U.S. military action within a reasonable distance and/or within a reasonable timeframe as that identified in the report; (2) a review of all available information, including information derived through intelligence sources, video from the weapon platform and/or ISR assets, and any information provided in the report, leads to the determination that it was more likely that civilians were not injured or killed; or (3) the report did not include enough information to assess it. 

As a result, this report, like its predecessor, fails to truly account for all “reasonably suspected” civilian casualties, because the department uses the very high “more likely than not” standard to assess a claim of casualties as “credible.” The effect is to place a number of casualty reports into a larger bucket of “non-credible” claims, when the department has sufficient certainty that DoD was at the scene but “insufficient evidence” of civilian casualties.

This issue of the evidentiary threshold required for an external allegation to achieve an assessment of “credible” thus remains an important area where there is a lot of room for improved policies and practices. 

Details about the contents of a new DoD policy and internal reforms: 

In a few parts of the report, the department provides hints as to what may be contained in a forthcoming DoD policy on civilian casualties, to include guidance for assessing and investigating casualties and a policy on ex gratia.

The report also includes a description of some internal developments that may aid in refining DoD’s approach to civilian casualties, including modifications to the Joint Instruction on Targeting with lessons learned from operations such as Inherent Resolve. The department also has established a civilian casualties working group, which was previewed in a separate report delivered in February in accordance with Section 936, which required DoD to place a senior official in charge of civilian casualties policy. 

Changes to “declared theater of armed conflict”: 

The new report also contains an interesting paragraph explaining that it considers all of the theaters it is reporting on it as falling within “a declared theater of active armed conflict.” As a result, the report’s breakdown of strike data does not include a column for indicating whether each individual strike occurred inside or outside such areas, as required by Section 1057. The department’s explanation says nothing about the fact that Section 1057 uses this parameter, so regardless of changes to other statutory provisions, it is still current law for purposes of this report:

Last year’s report used the term “a declared theater of active armed conflict,” as that term was understood in the context of Title 10 U.S.C. § 130f. Title 10 U.S.C. § 130f has since been amended and no longer includes the term “a declared theater of active armed conflict.” The term “a declared theater of active armed conflict” is also not defined in relevant DoD doctrine. For the purposes of this report, the term “a declared theater of active armed conflict” will be considered to mean, for calendar year 2018, Afghanistan, Iraq, Libya, Somalia, Syria, and Yemen. Thus, all U.S. military operations and particular instances listed below that resulted in civilian casualties occurred in a declared theater of active armed conflict, in the context of the ongoing armed conflict against al-Qa’ida, the Taliban, and associated forces, including ISIS.

What’s interesting, and concerning, about the lack of clarity on the correct and current terminology here is that there is currently little to no transparency about U.S. government strikes outside what the Obama administration dubbed “areas of active hostilities.” For instance, we do not know whether the 2013 Presidential Policy Guidance has been amended (aside from what we read in the New York Times), whether it still applies, and, if so, where it applies. Are the countries included in today’s report the only ones where the U.S. is conducting lethal strikes? Nor do we know how many people, however categorized, are being killed in such strikes. The current administration has not complied with, and has now officially revoked, the minimal reporting on strikes outside “areas of active hostilities” that was required by executive order but not by statute. 

Where DoD Should Go from Here 

Importantly, the report, along with the civilian casualties study, lays bare the need for DoD to be more proactive – and to find more innovative ways – in seeking out information that it may not have readily at its disposal. Intelligence, full motion video, and pre-strike information are all useful, but may not tell the whole story.

DoD also should carefully review its high threshold for determining whether a report is “credible,” especially when it has sufficient information to suggest the U.S. was involved in a strike but isn’t sure if anybody died or was injured as a result. Some reasonable modifications to procedure would make for more accurate figures overall and bring the numbers from outside organizations closer to those DoD itself maintains and reports. Increased accuracy is not only important for after-the-fact assessments, but for being able to better predict — and thereby better avoid — civilian casualties on the front end. (And, of course, there are many other issues to address as well, such as the ways DoD assesses who is targetable.)

Where Congress Should Go from Here 

It’s also clear there is more that Congress can do to strengthen use-of-force transparency, building on the very important provisions it has passed in the last two years on a bipartisan basis.

A wide range of organizations, including transparency, civil liberties, and religious organizations, as well as human rights organizations that conduct civilian casualties investigations in the field, sent a letter to the House and Senate Armed Services Committees today recommending several concrete steps that Congress should take to build on its recent legislative efforts.

In the letter, the signers, which include our respective organizations, Human Rights First and  Center for Civilians in Conflict, recommend, among other steps, that Congress:

  • require public transparency regarding any changes that have been made to the 2013 lethal targeting policy known as the Presidential Policy Guidance.
  • require public reporting of the names of all groups and countries the U.S. is using force against or in.
  • restore the reporting requirements in the July 2016 executive order on civilian casualties.

With the House and Senate both scheduled to begin marking up their respective defense authorization bills, now is the perfect time to take such steps.

(*The authors are grateful to Chris Woods of Airwars for assistance in compiling estimates made by tracking organizations.)

Image: Alex Wroblewski/Getty