Predicting Daily Feed Costs for Dairy Management Models

¹ Dairy Science Department, University of Florida, Gainesville 32611
² Food and Resource Economics Department, University of Florida, Gainesville 32611

Two constant nutrient value methods of predicting daily ration costs for management models were evaluated: constant dollars per megacalorie of NE_L (.050, .061, .072, and .095) and constant dollars per kilogram of TDN and CP (.132 and .265, respectively). The methods were compared with ration costs calculated by linear programming using high, average, and low corn price scenarios. Constant nutrient value methods were inaccurate predictors of ration costs, and their margin of error generally increased as milk yield and price increased. Deviations from ration costs calculated by linear programming were lowest for the low corn price; high dollars per megacalorie of NE_L (.072, .095) overestimated ration costs by $.014 to $1.017/d per cow, and low dollars per megacalorie of NE_L (.050, .061) underestimated ration costs by $.024 to $1.014/d per cow. Constant value per kilogram of TDN and CP overestimated ration costs by $.275 to $.579/d per cow. At the average price scenario, $.095/Mcal of NE_L and constant value per kilogram of TDN and CP overestimated ration costs by $.032 to $.456/d per cow, and lower dollars per megacalorie of NE_L (.050, .061, .072) underestimated ration costs by $.059 to $1.846/d per cow. At the high price scenario, both constant nutrient cost methods and all values underestimated ration costs by $.011 to $2.940/d per cow. Regression analysis of daily ration costs per kilogram of milk and per kilogram of DM indicated that estimation inaccuracies resulted from a failure of constant nutrient cost methods to account for a constant proportion of ration costs and increases in aggregate costs per unit of nutrient as milk yield increased. Alternative methods are suggested.

Key Words: feed costs • linear programming • shadow price

Submitted on December 9, 1991
Accepted on July 10, 1992