Landscape heterogeneity, basis risk and the feasibility of index insurance: an analysis of rice in upland regions of Southeast Asia

Despite its promise, the adoption of index insurance has been hindered by the extent of basis risk, the additional variability introduced by its reliance on a signal correlated with losses rather than losses themselves. We examine the feasibility of substantially reducing basis risk by accounting for heterogeneity in production conditions via clustering data into more homogeneous groups. We exemplify this approach using data from a sample of rice producers in northern Laos, using Normalized Difference Vegetation Index (NDVI) data as the index on which the contract is defined. Our results show that accounting for landscape heterogeneity substantially improves the insurance contracts that can be offered to rice producers.