Association for Public Policy Analysis & Management

The theory of poverty traps suggests that multiple equilibria welfare dynamics should emerge in the presence of multiple financial market failures and non-convex production technologies. Such dynamics have been observed empirically, most commonly in analyses where a single productive asset dominates the economy, such as among livestock herders in rural Kenya and Ethiopia. However, outside of such settings, multiple equilibria welfare dynamics are rarely identified. One explanation for this is that in most economies multiple production technologies are available, and the outer envelope of all these technologies may be convex. Alternatively, poverty traps may exist for heterogeneous subgroups, but not at the population mean, which is where current methods commonly look. Whether or not a poverty trap exists in a given setting, and for whom, is of great importance for policy and programmatic interventions. In the presence of poverty traps, intervention is not only merited, it is one of the only routes out of poverty. In the absence of poverty traps, intervention is not necessary and may even be harmful in the long run. In this paper, I develop a method for the examination of welfare dynamics in a setting where the livelihood strategy choice set is complex with the objectives of: 1) ascertaining whether the outer envelope of this choice set is convex and 2) observing whether heterogeneous welfare dynamics emerge from an analysis of these livelihood subgroups. I apply this method to a panel dataset from Kagera, Tanzania.