small share of costs (MacDonald and Ollinger, 2000 and 2005; Ollinger, MacDonald, and
Madison, 2005), the meat and materials term mainly reflects meat inputs.
The three factor model expressed in equation 2 is quite general with many possible
variations. A number of economists, such as MacDonald and Ollinger (2000, 2005), Antle
(2000), and Ollinger, MacDonald, and Madison (2005), faced a similar problem and used a
Gallant-Jorgenson (G-J) likelihood ratio test (a chi-square test) to choose the best models from
among sets of restrictive models. That same approach was followed here.
Table 1 gives a model number, description, test variables, and test, and the number of
parameters estimated and restrictions, and the G-J value and model chi-square for meat and
chicken slaughter and meat processing. Model testing was conducted in the following way. In
each industry, we began by comparing the most restrictive version of equation 2 containing
factor prices and output (P,LB) against least restrictive model (P,LB, T,S). Then, the least
restrictive model is compared against models with one variable excluded to evaluate the impact
of that one (removed) variable to model fit. Thus, in the first test, a base model consisting of
prices and output is compared against a model that also contains the technology index and
sanitation and process control effort. This test, a comparison of Model II with Model I,
indicates that technology use and performance of sanitation and process control tasks are jointly
significant in the meat slaughter and the meat processing industries but not in chicken slaughter.
The test of Model III versus Model II indicates that technology is significant only in meat
slaughter, and the test of Model IV versus Model II shows that sanitation and process control
effort were significant in both meat slaughter and meat processing. Neither technology nor effort
devoted to performance of sanitation and process control tasks were significant in poultry.
11