dc.description.abstract | Anecdotal evidence suggesting that feedlot cattle death loss is increasing over time has revived a need for further research on death loss in feedlots. Death loss may be related to controllable factors including technology adoption, feed rations, and cattle sourcing, as well as uncontrollable factors such as weather. This dissertation uses information from cattle feeding operations in the Southern Great Plains to understand the situation. The first study implements tests of structural change, including the cumulative sum (CUSUM test), CUSUM of squares test, and Bai and Perron tests, on a basic model of steer death loss rate. The basic model is estimated using a monthly observations where death loss rate is a function of in-weight, days on feed, time trend, and monthly dummies. The results suggest that there is structural change during the period of December 2000 – September 2010, which is also supported by a test of unequal means and variances. The second study use a Tobit model to estimate feedlot death loss rate using pen-level data. A Tobit model is used because not every observed pen has dead cattle which censors the data at zero. In addition to in-weight, pen size, sick head days, and cattle treatments, the results found that cattle source, geographic location, and market source type are important determinants for death loss. In the third study, expected net returns are computed based on a recursive system model for feedlot. The expected net returns take into account the expected death loss rate from the second study. Twelve scenarios for steers with a mix of in-weights, pen sizes, closeout months, and market origins are created. The results suggest that heavier in-weights and country-sourced (directly sourced from ranches) cattle yield greater returns, depending on the closeout months. Net returns for heavier in-weights are also found to be very sensitive to corn price changes. | |