The welfare and behavioural impact of group-based insurance provision in Ethiopia

In the presence of missing insurance markets in rural Africa, indigenous risk sharing groups have been able to provide insurance to their members against some types of risk. Based on unique data on 78 funeral insurance groups and 300 households from rural Ethiopia, a first background paper has described and modelled the welfare impact of semiformal insurance groups both theoretically and empirically (Bold and Dercon, 2009). In particular, we find that  funeral associations - known as Iddir in Ethiopia - are large and sophisticated institutions, which are formed with the specific purpose of offering insurance for funeral expenses and increasingly other forms of insurance and credit to cope with hardship, which  are much larger than the informal insurance networks usually considered in the literature, but still considerably smaller than the community. Groups are also distinguished by the type of contract they offer. There are a large number of groups that demand regular (ex-ante) contributions and accumulate substantial savings and these exist alongside more informal associations that collect contributions only at the time of a funeral (i.e. ex-post) and hold few if any savings; in other words, groups differ in terms of their institutional design.

Our paper explores the environmental constraints that lead these two contracts to coexist when a contract that uses savings clearly dominates a contract without this feature in a first-best world. We find that imperfect enforceability which limits the extent of stable group formation and differential levels of social capital and trust explain the heterogeneity of contracts. Based on our theoretical model, we predict that only if the extent of risk-sharing is limited by the threat of coalitional deviations, would we expect to observe heterogeneity in contracts. In that case, observed groups with savings would outperform groups without savings in terms of consumption smoothing. Furthermore, selection into contracts would be governed by the maximal regular contribution that is feasible in the group, which could be understood as either a monetary transfer or a credible threat to a social sanction. As a result, we expect groups with savings to be groups with a higher ability to raise funds or higher social cohesion (such as in the form of trust or other social connectedness). If the ability to save more is also related to issues of trust (avoiding embezzlement), then this prediction is even stronger.

We show in the data that the type of contract chosen by groups is correlated with the level of trust and other enforcement improving factors. We also predict that among the observed contracts, those with group-based savings and ex-ante payments will attain higher welfare in terms of consumption smoothing than those observed using no group savings. Using panel data, and controlling for household fixed effects and time-varying village level fixed effects, we show that funeral groups are vehicles for risk-sharing and that contract type matters for performance in line with these predictions. In particular, we confirm that groups with contracts including group savings resulted in welfare gains in the form of higher consumption smoothing among the members of groups engaging in this type of contract, compared to those without group savings and only ex-post payments.

Despite this heterogeneity, our results suggest nevertheless that funeral associations in rural Ethiopia are remarkably successful in providing insurance against death shocks. In a broader context, the work also contributes to the extensive and continuously growing literature on social capital and its welfare impacts. While high levels of trust and reciprocity are a necessary prerequisite for the enforcement of more sophisticated contractual arrangements, they are neither sufficient nor are they substitutes for the benefits derived from more formalized insurance provision.

These results and observations motivate further research currently in progress that examines how these insurance groups can be scaled up both in terms of size to insure their members against other types of risk such as health shocks or weather risks. There is widespread evidence that uptake of especially weather insurance products at the individual level is fairly low. A team led by Ruth Vargas-Hill (IFPRI) has conducted a detailed survey in rural Ethiopia examining financial literacy and willingness to pay at the individual level both via direct survey questions as well as insurance and risk-aversion games. Moreover, the survey has explored whether funeral insurance groups (whose leaders have more extensive experience and knowledge of insurance and who are trusted individuals within the community) may be appropriate institutions for rolling out health and weather insurance products. The data is currently being analysed with encouraging results. On the basis of this we are currently in the process of designing appropriate insurance products. The ultimate aim is to roll these out as part of a randomized controlled trial in which both individuals and insurance groups are targeted and to compare uptake and welfare effects of these programs.