Stochastic Frontier Yield Function Analysis to Predict Returns to a New Crop: An Example of Camelina Sativa Yields Conditional on Local Factor Levels

The purpose of this study is to develop a model that calculates the probability distribution of camelina expected yields dependent on location-related variables such as precipitation, temperature, and solar radiation, as well as nitrogen rate and others. Camelina is an oilseed crop grown in cool climate with low input requirements including little water. The application to camelina addresses challenges in analysis of potential adoption of crops with limited field data. Our data include trials and crop yields in the United States from 2005 to 2012. They have been assembled from various published reports covering a range of locations, seasons, and production methods. We begin by fitting a least squares (LS) regression model to camelina yields. As a robustness check we also apply a stochastic frontier framework under Cobb-Douglas technology. Preliminary results indicate that the average maximum precipitation for the period of interest positively affected the mean camelina yields, whereas it has no impact on yield variability. An increase in average maximum precipitation will more likely decrease the technical inefficiency. Both higher nitrogen rates and higher average maximum growing degree days will more likely increase the average yields. A taller camelina plant positively affects the mean yields and the yield variability. In contrast, total solar radiation is negatively correlated with mean yields and variation. There is still much to be learned about the crop and its best management practices as production expands. The analysis of the interaction of managed input variables and environmental factors will help us assess varietal performance and provide location conditional predictions.