Correct pricing of agricultural insurance products designed to hedge against the risk of extended dry spells depends on knowledge of the relevant precipitation statistics, as well as a stationarity assumption. In some regions, however, instrumental records are short, incomplete or absent, rendering estimation problematic, while at the same time climatic change may render the background state nonstationary. Here, an evaluation is first made of the utility of tree-ring records in constraining precipitation variability in the Central American region. Records from a variety of locations are considered, and the sensitivity of pricing structure to record length and quality assessed. An investigation is then conducted of the stability of the derived relationships under a set of defined climate change scenarios.