Chatham House Report on Proportionality in the Conduct of Hostilities — Some Key Elements

A- A A+

Cross-posted from EJIL:Talk!

At the end of 2018, the International Law Program at Chatham House published a report analyzing the key steps in making assessments about proportionality under international humanitarian law, with a particular focus on incidental harm. The rule of proportionality as formulated in Article 51 of Additional Protocol I of 1977 (AP I) requires belligerents to refrain from attacks ‘which may be expected to cause incidental loss of civilian life, injury to civilians, damage to civilian objects, or a combination thereof, which would be excessive in relation to the concrete and direct military advantage anticipated.’

The report addresses three sets of questions: first, the criteria of causation and foreseeability, the weight to be assigned to particular kinds of harm, and how to assess whether the expected incidental harm is excessive in relation to the anticipated military advantage; second, the types of incidental harm to be considered in proportionality assessments; and, third, a number of legal questions raised by the implementation of the rule in practice.

After putting the rule of proportionality into context, this post presents four of the points that the report seeks to clarify. These are: the causation of the incidental harm and whether it is foreseeable; ‘knock-on’ or ‘reverberating’ harm; the types of injury to civilians to be considered; and the notion of ‘excessive’ incidental harm.

Putting the rule of proportionality in context and the notion of ‘attack as a whole’

The rule on proportionality must not be considered in isolation. It forms part of a framework of rules that gives effect to the general obligation in the conduct of military operations to take constant care to spare civilians and civilian objects. These rules include the principle of distinction; the prohibitions on directing attacks against civilians and civilian objects, and on indiscriminate attacks; and obligations relating to precautions in attack and against the effects of attacks.

Not every death or injury to civilians or damage to civilian objects in armed conflict must necessarily be assessed through the lens of the rule of proportionality. For the rule to come into play, a number of conditions need to be met. First, the harm must be incidental – that is, it must occur in the course of an attack directed against a military objective. Second, the harm must be expected to arise as a result of ‘an attack’ as this term is understood in IHL. Not every use of force or military operation in an armed conflict constitutes ‘an attack.’ There is no common understanding of what constitutes an ‘attack.’ The Group of Eminent Experts on Yemen recently took an extremely broad approach. Finally, the rule requires weighing the incidental harm expected from a specific attack against the military advantage anticipated from that same attack.

Article 49 AP I defines an attack as ‘acts of violence against the adversary, whether in offence or in defense.’ States have emphasized that for the purposes of proportionality, it is the military advantage anticipated from ‘the attack considered as a whole and not only from isolated parts thereof’ that must be considered. This is so even though such isolated or particular parts may constitute distinct ‘attacks’ as defined in Article 49 AP I and, consequently, have to comply with other rules governing the conduct of hostilities, starting from the requirement that they be directed against military objectives.

In determining what amounts to an ‘attack as a whole’ and, therefore, constitutes the frame of reference for conducting a proportionality assessment, consideration must be given to the contextin which the act is conducted. If the military advantage anticipated from a single attack is not dependent on or affected by other acts, then the act should be considered an ‘attack as a whole’ for the purpose of proportionality assessments. If, on the other hand, a single attack is an element in a larger operation where other acts contribute to the military advantage, then the operation in its entirety should be considered the ‘attack as a whole.’

Whatever the precise parameters of ‘an attack as a whole’ in a given case, it is essential that the same interpretation and, thus, frame of reference be adopted for determining what falls within the two sides of the proportionality assessment: that is, in the estimation of both anticipated military advantage and of expected incidental harm. Doing otherwise would undermine the balancing exercise at the heart of the rule.

Key elements of proportionality assessments: causation and foreseeability

Two distinct issues must be considered in identifying the incidental harm to be taken into account in proportionality assessments: causation and foreseeability. Can the expected harm be considered as caused by the attack; and could that harm have been expected (that is, was it foreseeable) when the attack was launched? As proportionality requires an ex ante assessment before the attack is conducted, the relationship between causation and foreseeability is extremely close. These nonetheless remain two distinct elements in the analysis and must be considered separately.

Causation requires establishing a relation between an attack and particular incidental harm. If the attack is expected to be the sole cause of the harm, determining causation is straightforward; so much so that in practice the proportionality analysis focuses solely on foreseeability. In such situations, incidental harm can be considered as caused by the attack if it is the outcome that was expected to occur from the attack in the ordinary course of events.

Establishing causation becomes more complicated when more than one actor is involved – this could be the opponent or some other actor that is not a party to the conflict, such as an intergovernmental organization that imposes sanctions. Setting aside questions of foreseeability for now, the fact that a second actor has intervened does not automatically preclude the ensuing incidental harm from being considered as caused by the attack.

The incidental harm to be considered is that which would not occur but for the attack – with one limitation. Harm that results from the conduct of an actor other than the one carrying out the attack, and that does not arise from the physical effects of the attack, is excluded. By way of example, the death or injury of persons used as involuntary human shields is included in the incidental harm to be considered. Even though it is the conduct of the attacker’s opponent that has placed them at risk, the harm is a result of the physical effects of the attack. On the other hand, if the opponent were to execute people in retaliation for an attack, the harm would not be considered as caused by the attack as it was not caused by the physical effects of the attack.

The second element in the analysis is foreseeability. Not only is it necessary for the incidental harm to be caused by the attack; its occurrence must also have been foreseeable at the time the attack was planned or launched. In other areas of public international law, ‘foreseeability’ has been interpreted as referring to reasonable foreseeability: i.e. what should have been foreseen by the party responsible for the wrongful act at the time of the act. The same approach applies to the identification of the harm to be considered in proportionality assessments. It is that incidental harm that a reasonable person in the place of the person planning or launching the attack should have foreseen.

What can ‘reasonably’ be foreseen depends on the circumstances in which the attack is planned, decided or launched. This includes, in particular, the attacker’s capabilities and available resources; whether the attack was part of a pre-planned operation or occurred during dynamic targeting; and the context in which the attack was planned and conducted, including factors such as the time available, terrain, weather, capabilities, available troops and enemy activity.

‘Knock-on’ or ‘reverberating’ harm

The analytical framework just outlined facilitates consideration of ‘knock-on’ or ‘reverberating’ harm: incidental harm that does not arise immediately or in one causal step. Provided that the harm falls into one of the categories identified in Additional Protocol I, the geographic or temporal proximity of the harm to the attack is not determinative. Nor is the number of causal steps between the attack and the harm. While the anticipated military advantage to be considered in proportionality assessments must be ‘direct’, there is no such requirement for incidental harm. Instead, what matters is that the harm meets the criteria of causation and foreseeability. The incidental harm to be considered is that harm which would not occur but for the attack, but excluding harm that results from the conduct of another actor and is not due to the physical effects of the attack; and which was reasonably foreseeable at the time the attack was planned or launched, on the basis of information that the attacker had or could reasonably have been expected to have in the circumstances.

What constitutes incidental harm?

The rule of proportionality as formulated in Additional Protocol I lists three types of incidental harm to be considered: loss of civilian life, injury to civilians and damage to civilian objects. The Chatham House report focuses on these three types of harm without prejudice to the possibility that under the rule of proportionality under customary law (see Bartels, at p. 304), specific IHL protection regimes (see ICRC (2016), Commentary at paras 1353–1357, 1797 and 1987) or a ‘general principle’ of the law of armed conflict (see Kleffner, p. 43), other forms of incidental harm should also be considered.

What types of injury to civilians should be considered in proportionality assessments? There is a medical distinction between an ‘injury’, which is caused by the application of external forces, and a ‘disease.’ However, the negotiating history of Additional Protocol I does not suggest that it was the intention of the drafters to exclude diseases. There appears no reason to adopt a narrow interpretation of the types of physical harm to be considered in proportionality assessments. Excluding disease could give rise to absurd results, for example, requiring expected injuries caused by the blast of nuclear weapons to be considered, but not those caused by the exposure to radiation from the same attack. The severity of the expected harm affects the weight to be assigned to it in proportionality assessments.

What about mental harm? IHL prohibits the intentional infliction of certain types of mental harm, including acts or threats of violence whose primary purpose is to spread terror among the civilian population. These prohibitions are set out in the Additional Protocols of 1977. The fact that provisions referring explicitly to mental harm are to be found in the same instrument as the rule of proportionality, or in an instrument negotiated in the same diplomatic process, should not be interpreted as indicating that their omission from the rule means that the drafters intentionally excluded mental harm. Additional Protocol I does not define ‘injury’, and the fact that mental harm was not addressed during the negotiations of Articles 51 and 57 AP I is probably merely a reflection of the fact that at the time it was not considered that attacks, unlike torture, could also cause mental harm.

Taking this into account, and in view also of the developments in the medical understanding of mental harm since 1977, and its increasing recognition in human rights law and IHL, there is no reason in principle to exclude mental harm from the scope of proportionality assessments. That said, the nature of mental harm raises a number of challenges to key elements of proportionality assessments, including questions of causation, foreseeability and the weight to be assigned to the risk of the occurrence of mental harm.

In terms of causation, the rule of proportionality requires identification of the incidental harm that can be expected as a result of a specific attack. While this is usually straightforward for physical injury, doing so with regard to mental harm is more complex. Civilians are frequently exposed to hostilities for prolonged periods, making it difficult to determine whether a particular attack is likely to cause mental harm.

With regard to foreseeability, the occurrence of mental harm is more subjective than that of physical harm. This makes it more difficult to foresee. The effects on mental health of exposure to an attack are likely to vary significantly, both among individuals, and also depending on the context where the attack occurs.

While concerns have also been expressed about the difficulties of quantifying mental harm – that is, assigning a weight to its occurrence. It is unclear why this should be more difficult than for other types of harm. On the contrary, it is precisely at this weighing stage of proportionality assessments that some of the difficulties currently raised by mental harm can be addressed. These include the likelihood of it occurring and its severity.

When is incidental harm ‘excessive’?

The rule of proportionality prohibits attacks expected to cause incidental harm that would be excessive in relation to the anticipated concrete and direct military advantage. Neither Additional Protocol I nor military manuals provide guidance on how to interpret the notion of ‘excessive.’ There is no indication that the imbalance between expected incidental harm and military necessity needs to be significant for the rule to be violated.

Determining whether the expected incidental harm would be excessive is probably the most challenging aspect of the application of the rule of proportionality in practice. It requires valuing and comparing two incommensurable factors: military advantage and incidental harm. This said, proportionality in other areas of law frequently requires this type of assessment. For example, in human rights law proportionality can require balancing restrictions of certain rights with considerations of national security (see Article 8(2) ECHR). Moreover, it is an assessment that military commanders are constantly undertaking, so while it may be difficult to attempt to set parameters, it is not an impossible determination to make in practice.

‘Excessive’ is a wide but not indeterminate standard. Provided they do what is required to collect information on which to base their assessment and conduct the assessment in good faith and in a manner that is reasonable, belligerents have ‘a fairly broad margin of judgment’ (see ICRC Commentary to the Additional Protocols, para 2210), to determine whether the expected incidental harm would be excessive.

Members of the armed forces consulted for the Chatham House report were unable to share examples of instances when expected incidental harm would be excessive. Not only was this classified information, but frequently belligerents set themselves stricter obligations as a matter of policy than those required by law. These may preclude attacks in circumstances in which the incidental harm would not necessarily be excessive as a matter of law.


Assessing compliance with the rule of proportionality is notoriously difficult. Those attempting to do so frequently do not know what the anticipated military advantage is, and the notion of ‘excessive’ gives those planning or deciding an attack a considerable – but not unlimited – margin of judgment.  This does not mean that the rule serves no purpose in sparing civilians from the effects of hostilities.

It imposes important obligations on belligerents to establish systems to gather and analyse relevant information and to ensure that this information is taken into account in the targeting cycle; to conduct ‘lessons learned’ processes as soon as possible after attacks in order to inform future attacks; and to conduct assessments of attacks when the rule may have been violated. As noted by representatives of armed forces involved in the elaboration of the Chatham House report the challenge lies in integrating these elements into their targeting practices.