Modeling Crop Yield Distributions from Small Samples

Accurately modeling crop yield distributions is important for estimation of crop insurance premiums and farm risk-management decisions. A major challenge in the modeling has been due to small sample size. This study evaluated potentials of L-moments, a recent concept in mathematical statistics, in modeling crop yield distribution. Five candidate distributions were ranked for describing the wheat yields. The selected distribution was robust for small sample and was invariant to de-trending. The result was consistent with that from the maximum likelihood and goodness-of-fit method.