The objective of this study is to evaluate the risk associated with major agricultural commodity yields in the United States. We are particularly concerned with the nonstationary nature of the yield distribution, which arises primarily as a result of technological progress and changing environmental conditions over time. In contrast to common two-stage methods, we propose an alternative parametric model that allows the moments of yield distributions to change with time. Several model selection techniques suggest the proposed time-varying model outperforms more conventional models in terms of in-sample goodness-of-fit, out-of-sample predictive power, and the prediction accuracy of insurance premium rates.