A growing literature on poverty traps advocates that social protection policies which aim to help households manage risk will more effectively alter poverty dynamics if such policies also account for a critical threshold, around which both equilibrium outcomes and optimal behavior bifurcate. In this paper, we account for this type of threshold, and examine whether insurance can achieve the goals of social protection. Unlike traditional publicly-provided social protection, such as cash transfers or food aid, insurance is a market-based risk management tool that protects households only if they self-select into purchasing an insurance contract.

Stochastic dynamic programming methods reveal low demand for asset-protection insurance is optimal for vulnerable households whose assets are near a critical threshold, because the opportunity cost of insurance at the threshold is especially high. Paradoxically, these same households have the most to gain from protection of this kind. Despite low demand, an insurance-induced behavioral response and the ability to protect assets in the future reduces vulnerability for these same households. We consider this impact on poverty dynamics in a typical setting by calibrating the model to the northern Kenyan rangelands, where evidence of a poverty trap exists and pastoralists have the opportunity to insure livestock against drought losses.