Nonparametric Estimation of Crop Yield Distributions: A Panel Data Approach

We propose a flexible nonparametric density estimator for panel data. One possible areas of application is estimation of crop yield distributions whose data tend to be short panels from many geographical units. Taking into account the panel structure of the data can likely improve the efficiency of the estimation when the crop distributions share some common futures over time and cross-sectionally. We apply this method to estimate annual average crop yields of 99 Iowa counties. The results demonstrate the usefulness of the proposed method to estimate simultaneously densities from a large number of cross-sectional units.