Risk sharing and social hierarchy in disaster aid

A common notion is that communities can play an active role in identifying vulnerabilities to natural disasters, mitigating them and responding to them in proactive ways, and thus communities need to be given decision-making power. In order to design effective participatory/decentralised disaster management, it is crucial to better understand community decision-making mechanisms. A recently published paper explores a critical question about disaster aid: once aid is allocated to a community, how are provisions allocated among households within the community? This question has not been sufficiently answered, especially in developing countries because of the paucity of household-level data on disaster damage and aid, with rich variations within communities. With such data, the extent to which targeting within the community is effective can be directly measured by aid response to damage in a regression model controlling for the community-level aid supply.

The author’s original post-disaster surveys in rural Fiji collected such micro-level data²; Fiji’s disaster management is strongly centralised; in particular, communities do not participate in decisions about aid allocation across communities. Under the status quo, however, each community can influence how aid is allocated among its members, because local governments and aid agencies, with limited capacity for assessing household-level damage and implementing intra-community allocations, rely on community resources in practice and they are linked with those who have power within the communities.

The paper explores two community mechanisms: informal risk-sharing institutions and social hierarchies. First, extensive economic studies show that with a lack of formal insurance, informal risk sharing – reciprocal help based on implicit contracts – plays a central role as a safety net in poor populations. Second, a social hierarchy can strongly shape a community’s political economy, which determines local benefit allocation as highlighted in community-based development. Then, elite domination can be a secondary driving force of aid allocation. For given damage, elites are more likely to be recipients or to receive a greater amount than non-elites. This elite domination for given damage determines the equity of aid allocation. The limited observability of the process of benefit allocation within communities, however, is a common problem in community-based programmes and is especially serious in post-disaster contexts.  The paper overcomes this challenge by exploiting direct measures of social status. As the author’s survey stratified households in each village by their kin group and elite status, direct measures of elite status at the household and kin group levels with rich variations are available. With the author’s micro-level data of social status, elite domination for given damage can be measured by aid response to social status in a regression model controlling for household-level damage and unobserved village heterogeneity, such as the process of aid allocation.

Treating risk sharing and social hierarchy independently, the paper shows evidence for both seemingly weak targeting and elite domination for given damage. Building on kinship, these two community mechanisms are not independent of each other; in particular, kin groups are a major village subgroup in both. The paper reveals that the interaction of the two mechanisms can lead to elite domination in targeting on damage; aid allocation more strongly responds to elite’s damage than non-elites. The analysis compares emergency food aid in the relief and early recovery phases and housing aid in the recovery/reconstruction phase.  It also compares their allocations within the kin group and across kin groups, as well as allocations of food aid on housing and crop damage. As such, the paper demonstrates how patterns of targeting and elite domination vary over post-disaster phases, at different levels of allocation and across different shocks. Specifically, disaster victims are allocated less food aid in the early post-disaster phase, mainly because they receive greater labour help for housing rehabilitation, especially within kin groups; in contrast, the allocation of housing aid in the late phase is strongly targeted on housing damage.  There was also evidence for potential elite capture; for given damage, local elites dominate the allocation of housing aid but not food aid.

The paper’s framework and findings can apply to contexts where a natural disaster with moderate severity does not undermine social systems (hierarchy) and the community’s risk sharing works between disaster victims and non-victims. Such natural disasters are common across developing areas and effective management to cope with them in a self-reliant manner, e.g. through a community-based approach, is of great importance. If instead, all or most community members suffer an unusual catastrophic hazard, such as the 2004 Indian Ocean Tsunami, not only does the risk-sharing institution not work, thus making the issue of seemingly weak targeting irrelevant, but also social systems per se can be damaged through massive casualties, evacuations and migrations. 

These results lead to the following implications for local disaster management in developing countries:

1. For better allocation of disaster aid within communities, private risk-sharing institutions should be maintained and strengthened as a self-targeting mechanism. Participatory community-based disaster management needs to be integrated with broad, community-based development programmes (e.g. poverty alleviation).
2. Although targeting aid on damage is critically important, the performance of overall risk sharing needs to be given direct attention. This is especially so in the early post-disaster phase when private risk sharing potentially makes up limited aid; however, information about it is lacking the most at that time. That is why strengthening local institutions before the event in a self-reliant manner is strongly demanded.
3. In hierarchical developing societies, elites and/or elite groups can dominate intra-community aid allocation not only for given damage, but also for targeting on damage throughout the post-disaster phases. More attention should be given to the political economy of local disaster management. Although its limited observability is a serious constraint, micro-level data of social status can help overcome this challenge.
4. The coexistence of risk sharing and social hierarchy as allocation mechanisms poses a great challenge for policy makers, because the former is often built on the latter, especially through kinship. On one hand, interventions to neutralise elite capture (external accountability mechanisms) may weaken local safety nets instead of strengthen them; on the other hand, when such interventions are weak or lacking, equity can be significantly undermined. To balance this trade off, a better understanding of the process of local aid allocation under the status quo is crucial, as in local benefit allocation of community-based programmes in general.

Footnotes

¹Takasaki. Y (2014). How is disaster aid allocated within poor communities? Risk sharing and social hierarchy. Journal of International Development, vol 26, pp 1097-1114 (2014). DOI: 10.1002/jid.2985

²On the 13th January 2003, Cyclone Ami swept over the northern and eastern regions of the Fiji islands. The author conducted two rounds of surveys; between August and November 2003 among 374 households in nine villages and between July and September 2005 among 906 households in 43 villages.