Conference on Applied Statistics in Agriculture

ESTIMATING THE CONTRIBUTION OF SPECIFIC WHEAT VARIETIES TO STATE-LEVEL PRODUCTION

This paper addresses measurement issues involved in estimating the benefits from a university wheat breeding program. The conceptual foundation is to estimate the spatial distribution of farm-level yields for varieties in each year, based on Nebraska winter wheat variety test data from 1972 to 2001, with 9-20 locations per year. We can then estimate the state average yield difference that would occur between the portfolios of varieties that farmers actually grew, and the portfolio of varieties excluding those varieties from the university breeding program. This paper reports estimates of the statewide annual yields of individual varieties using a standard fixed effect model, and with anisotropic power as spatial covariance structure. The results indicates that UNL-developed varieties added significant production increases until the mid 1990's when their portfolio advantage diminished due to the introduction of several high yield private varieties. The lump sum direct producer benefit of the Nebraska varieties is estimated to be about $7 million per year in Nebraska during the 1972-2001 period. Incorporating the spatial structure on variety test data, our estimated producer benefit are about 20% higher than if spatial structure had been ignored.