Risk pooling and limited commitment: an experimental analysis

This paper presents rigorous and direct tests of two assumptions relating to limited commitment and asymmetric information that underpin current models of risk pooling. A specially designed economic experiment involving 678 subjects across 23 Zimbabwean villages is used to solve the problems of observability and quantification that have frustrated previous attempts to conduct such tests. I find that more extrinsic commitment is associated with more risk pooling, but that more information is associated with less risk pooling. The first of these results accords with our expectations and assumptions. The second does not. I offer two explanations as to the origin of the second result and discuss their implications for how we view the assumptions made elsewhere in the literature. I also conduct a test of the relevance or external validity of the experimental results to our understanding of real risk pooling behaviour. In four out of the five villages for which the test could be conducted the networks of risk pooling contracts constructed during the experiment and the networks existing in real life were significantly correlated.