Milked for All They Are Worth:
Livestock Replacement in a Dynamic Discrete Choice Model

Job Market Paper

Click here for a copy of the working paper

Dairy farmers in the United States routinely cull animals before asset replacement models claim is profit maximizing. This paper examines cow replacement decisions on over 1,000 Wisconsin dairy farms during the period 2011-2014 to discern whether unexpected cow mortality drives replacement decisions. Since animal replacements must be procured ten months in advance in dairy, unexpected asset failure can incur large costs on dairy farms and may encourage early replacement before animal health declines. I model the choice as a dynamic discrete choice problem and estimate the model parameters while taking into account unobserved, fixed cow heterogeneity. Using the conditional choice probability method paired with machine learning, I estimate the cost of mortality at 2,300 USD per death. This is more than twice what is calculated from simulations, which suggests dairy cow replacement models are at odds with producer behavior because they have underestimated the costs of declining animal health. Utilizing farm size heterogeneity, I also find that mortality costs are three times higher on small dairies than on larger ones. Though dairy breeders have prioritized production over health in past decades, these estimates suggest that breeding instead for health and longevity can generate significant cost savings for dairy producers.