Quantifying Heterogeneous Returns to Genetic Selection: Evidence from Wisconsin Dairies
Department of Agricultural & Applied Economics
University of Wisconsin - Madison
Wednesday, October 31, 2018
Taylor-Hibbard Seminar Room (Rm103)
12:00 pm-1:30 pm
In the agriculture sector, a large vector of productivity growth has been improvements in biological innovations, specifically improvements in genetics (Olmstead and Rhode 2008). The dairy sector in the US is an interesting case study for such growth, which has been achieved by innovations in the delivery of genetics, via artificial insemination, and innovations in institutions and data collection, via herd testing associations. The current system of evaluation uses on-farm data to evaluate the productivity of dairy bulls, which provides amazing scale for evaluation but may lead to biased estimates of productivity. Specifically, a related literature in technology adoption suggests that farms may adopt bulls they already have an advantage in using. In this paper, we ask whether the current industry estimates of mean productivity are 1) reliable, that is whether selection behavior biases mean returns, and 2) useful, that is how heterogeneous are returns across farms? Using a data set of dairy bull evaluations and cow level production data, we determine groups of sire genetics using K-means clustering and estimate their mean returns as well as their returns across herds using variation in selection behavior rather than relationships between sires. Our preliminary results show a large amount of heterogeneity in returns and the potential for variation in genetic merit values across sires to correct bias in mean estimates.