Citizens' aid preferences amid state collapse: Experimental evidence from Haiti

Abstract Motivation Citizens' preferences signal which intermediaries they view as legitimate, trustworthy, and responsive, as well as capable of guiding donor decisions about aid‐delivery channels. However, little is known about these preferences in extremely fragile environments, where aid is both the most urgently needed and the most difficult to deliver. Purpose This study examines whom citizens believe should deliver reconstruction and relief aid in extreme fragility situations, how these attitudes shift when state authority collapses, and whether preferences diverge across socioeconomic groups. Methods The analysis is based on an original online survey conducted in Haiti (N = 2,001) in winter 2023/2024. Respondents evaluated how strongly they supported aid delivery through different actors in a vignette experiment. While fielding the survey, gangs launched a revolt which triggered the collapse of the central government. This created a natural experiment to compare attitudes before and after institutional breakdown. Findings State collapse reshaped citizen preferences, but not as most of the literature would have predicted. Citizens did not gravitate toward foreign NGOs, whose legitimacy in Haiti is weakened by well‐documented abuses and programming that is disconnected from local needs. Instead, the most significant shifts concerned local actors, especially municipalities, who are no less corrupt than the central government. Importantly, affluent and non‐affluent respondents moved in opposite directions. The affluent preferred channels that align with prevailing donor practices, while the non‐affluent—those most likely to be aid recipients—shifted toward municipalities. Policy implications Findings challenge the common donor practice of routing aid through external, non‐state channels. In contexts with a history of aid‐related harm, this risks further eroding trust. However, acting on citizens' signals requires care: if donors respond to the most audible local voices, they risk aligning aid allocation with the preferences of more affluent groups rather than those of intended beneficiaries. Municipal channels, while preferred by the non‐affluent, remain vulnerable to capture by elites or armed groups. In conflict settings, stated preferences may reflect strategic accommodation rather than genuine endorsement. While the focus is on Haiti, similar dilemmas could be unfolding across other fragile, conflict‐affected settings.