Risk-sharing and food aid in Ethiopia.

Part of the work in this area has focused on standard safety net policies and the implications of the presence of informal insurance systems. The issue is that targeting has very different effects depending on whether there is informal risk sharing or not. For example, targeted transfers to specific members of a risk-sharing network has no effect if they form a perfect risk sharing group. If the risk-sharing group is constrained by enforcement problems of perfect risk sharing, then these transfers may even have negative effects on those not receiving transfers. Poor targeting, on the other hand, would not be a serious problem if there are strong risk-sharing networks – they will share the transfers anyway.

Focusing on food aid in Ethiopia it is indeed found that some risk-sharing is present at the village level, including sharing of food aid. At the same time, targeting is relatively poor – with the poorest groups not necessarily receiving support. The result is, however, that despite poor targeting, the overall effect of food aid is beneficial for these poor groups, thanks to informal insurance arrangements. There is, however, also some evidence of crowding-out of other forms of insurance due to the presence of food aid: in villages with food aid, we find less evidence of informal support against idiosyncratic risk. The conclusion is that to assess the impact of safety nets, we need to take into account the presence and functioning of informal systems.