This thesis examines the approaches taken by the International Committee of the Red Cross (ICRC) and the Office of the High Commissioner for Refugees (UNHCR) to the protection of civilians during internal armed conflict, both at the level of global policy and at the level of implementation in the Colombian context. The thesis explains how the ICRC and UNHCR approach protection, why each has adopted its particular approach, and how and why the effectiveness of each approach is limited. In doing so, it offers a theoretical framework for explaining the approaches taken by international organizations (IOs) to new tasks within their mandates as well as policy implications for the ICRC, UNHCR and other humanitarian agencies. From a theoretical perspective, this research shows that factors internal to the IO carry greater explanatory power than external factors. Most significantly, when an IO expands into a new issue-area, it frames the new task in terms of the existing tasks within its mandate, replicating the specific goals and the means of pursuing those goals. The extent to which the approach is then adapted to the specificities of the new issue-area depends on the ‘bureaucratic personality’ of the IO, and specifically the extent to which decisions are informed by field-level experience. Internal conflicts by definition include armed non-state actors, and the analysis in this thesis emphasises both their significance in determining civilian security and their neglect in existing approaches to protection. While the ICRC seeks to reduce the threat posed by all armed actors (state and non-state) in its work at the field level, it relies heavily on an international legal framework which prioritises states and this partially undermines its attention to non-state actors at the field level. UNHCR retains a state-centric focus at both the field level and the level of global policy. From a policy perspective, therefore, the thesis advocates greater attention to armed non-state actors both at the level of practice and in the development of protection norms.

Cette thèse porte sur le contrôle que les Etats exercent sur les ONG humanitaires. Ces dernièresconservent parfois une grande marge de manoeuvre dans la mise en oeuvre de leur mandat, alorsque, dans d'autres circonstances, elles agissent en sous-contractants de la politique étrangère deleur État d'origine. Nous cherchons à identifier les facteurs qui mènent les ONG plutôt dans unedirection que dans l’autre, ainsi que la modalité de contrôle exercée par les États – coordinationou blanchiment - La thèse explique pourquoi, au sortir de la guerre froide, les Etats ont accru leurengagement dans l’action humanitaire, de façon unilatérale d’abord et multilatérale ensuite.Sur la base de l’étude du cas de la politique humanitaire internationale entre 1989 et 2005,l’analyse révèle un processus causal constitué de cinq séquences, qui explique comment les Etatsont recours à un agent quand leurs stratégies unilatérales sont infructueuses.Nos résultats attestent du caractère stratégique du financement de l’action humanitaire pour lesEtats, conditionné par les préférences géographiques et politiques de ces derniers. Financerl’action humanitaire permet de soutenir les anciennes colonies, d’appuyer la lutte contre leterrorisme et les interventions militaires des pays donateurs. Les financements étatiquesbénéficient aux ONG les plus dépendantes de leurs Etats d’origine. Les organisations nongouvernementalesdeviennent des organisations néo-gouvernementales. Celles-ci connaissent un rejetplus fort de leur présence en zone de conflit par rapport aux ONG qui refusent de recourir auxfonds publics. Face à l’échec de leurs stratégies de contrôle unilatéral, les Etats interventionnisteseuropéens délèguent la gestion de la politique humanitaire à un agent, ECHO, dont le mandat estde rendre moins visible la tutelle étatique les ONG.Cette situation engendre des problèmesd’agences multiples : l’agent dérape en raison de sa forte perméabilité à ses sous-contractants. Enréaction, les Etats renforcent les dispositifs de contrôle sur leur agent. Au final, la politiquehumanitaire se construit sur un équilibre fragile dans lequel chaque État tend naturellement à lacontrôler tout en sachant qu'un contrôle trop visible priverait cette politique de l'efficacité qui luiest propre, et qui tient, précisément à sa neutralité. / This thesis examines how States control their humanitarian NGOs. NGOs may sometimes enjoya great room of manoeuver in the implementation of their mandate, whereas, in othercircumstances, they act as sub-contractors of their home States’ foreign policies. This researchaims to identify the factors leading them to opt for one or the other course of action, as well as themodalities of States’ control (coordination or laundering). We explain why, at the end of the ColdWar, States have increased their commitments in humanitarian action, firstly bilaterally and then,multilaterally. Drawing upon the analysis of the international humanitarian policy between 1989and 2005, we reveal a 5-phases causal process which explain why States delegate competencies toan international agent when faced with the failures of their unilateral strategies. The results showthat States fund humanitarian aid in a strategic way, aligned with their geographic and politicalpreferences. Humanitarianism enables them to support their former colonies, to contribute toanti-terrorism policies, and to back military interventions. States’ funding mostly benefit to NGOthat depend on their home States. NGO thus become neo-governmental organizations. Theseagencies experience a greater rate of rejection of their action in conflict zones that NGOs thatrefuse to rely on public funding. Faced with the failure of their unilateral control strategies,European interventionist States delegate the management of the humanitarian policy to an agent,ECHO, who mandate is to make the State control of NGOs less visible. This situation raisesmultiple-agency problems: the agent slips because of its strong permeability to its sub-contractors.The principal react by reinforcing the control of its agent. At the end of the day, the humanitarianpolicy builds on a fragile equilibrium in which each State strives to control is while knowing thata too visible control decrease the efficacy of such policy, which precisely lies on its neutrality.