Three Essays on Modeling Risk in Fed Cattle Production

This study examines issues that arise when modeling risk associated with fed cattle production. While research concerning crop yield and revenue risks are numerous, studies focusing on production risk in livestock are much less frequent. The first essay evaluates the relationship among four variables associated with the health and performance of feedlot cattle, and the resulting production and profit risk. The four variables of interest include feed conversion rates, average daily gain, veterinary costs, and mortality rates, which are conditional on characteristics that are known when the pen is placed into a commercial feedlot. Conditional variables include gender, average weight, feedlot location, and season of placement. A multivariate Tobit model is used to characterize the relationship among the four dependent variables, where each element in the covariance matrix is conditional on placement characteristics. The second essay focuses on modeling cattle mortality rates, alone and as part of a system. A zero-inflated log-normal distribution is developed and shown to have advantages in model fit and prediction tests with this data set, relative to classical methods. A simulated data set is also utilized to assess the potential bias from assuming a Tobit model when the data are more accurately characterized through a mixture model. The third essay quantifies the amount of risk inherent in cattle feedlot operations through the use of simulation techniques. More specifically, extit[ex-ante] profit risks are evaluated under scenarios that utilize varying levels of price protection through the use of forward contracts and the options market.