The Determination of Optimal Rations for Pigs Fed Separated Milk and Grain*

Abstract

This paper demonstrates a procedure for determining economically optimal rations which avoids problems that arise in conventional analysis of experimental results. It consists of maximising a suitably constrained profit per unit time objective function, itself formed from functions obtained from pig fattening experiments under special Australian conditions. The relevant functions‐for consumption of milk and wheat, time to slaughter and percentage lean meatare first demonstrated, followed by a brief discussion of results obtained. Specifcally, it is noted that optimal diets differ according to whether the criterion adopted is profit per pig or profit per pig per week, that dairy farmers fattening pigs should feed substantial quantities of grain, and that the optimal diets are very stable to changing prices. More generally, the procedure avoids the statistical difficulties of conventional analysis of experiments arising from the estimation of weight‐gain functions based on actual quantities of food consumed.

Estimation and Application of Gain Isoquants in Decision Rules for Swine Producers*

“…A second estimation problem arises because the animal controls the rate at which feed is consumed (Battese, 1968). The researcher can only dictate either how much the animal eats per day or the number of days in trial.…”

Estimation and Application of a Production Function in Decision Rules for Swine Producers1

“…Step 2 Calculate the residuals for each of the seven periods 0, = Gi -G,, where G, is the vector of predicted total gain from (1). Step 3 To estimate the transformatipn necessary to correct for autocorrelation :…”

“…Different average starting weights for the pens of hogs create estimation problems that are not so much statistical in nature as they are mechanical. Consider equation ( 1 ) again. If observations on one pen of animals started at an average weight that was Go pounds greater than those of a second pen, then a corresponding translation of the input axes occurs by unknown amounts CO and Po.…”

Problems in determining the best levels of essential nutrients in feedingstuffs

“…These studies have been based on arbitrarily selected response models, usually quadratic (e.g. Dent, 1964;Sharpe & Dent, 1966;Battese et al 1968) and the reliability of their recommendations depends on the implicit assumption that any two response models which fit the experimental data equally well will provide recommended nutrient levels which do not differ to any marked extent. Through a simple practical example on protein level in diets for growing pigs we examine how true this implicit assumption is by comparing the recommendations from two equally plausible response models.…”