product or may not produce the two or more products specified in the cost function, many
observations would have to be dropped if a multi-product translog cost function were used.
Economists (Allen and Liu, 1995) and many others in trucking and other transportation
studies and MacDonald and Ollinger (2000, 2005) and Ollinger, MacDonald, and Madison
(2005) in hog, cattle, and poultry slaughter analyses have accommodated multi-product plants
with a single output translog cost function in which a single output is specified with a vector of
output characteristics that describe that output. The advantage of this approach is that one model
can be specified for both the single- and multi-product plants that may co-exist in an industry.
A single output, three factor translog cost function is specified in equation 2. The
variables identified in equation 1 are included in the empirical model. Notice that there are no
variables specifying characteristics for secondary products, as included in MacDonald and
Ollinger (2000, 2005) and Ollinger, MacDonald, and Madison (2005). Models with
characteristics were tested but they were dropped because they were not significant to model fit.
That left equation (2) with prices, output, food safety technology use, and effort devoted to
performing sanitation and process control tasks..
2
(2)ln C = αo + ∑ βi In Pi + ⅜∑ ∑ βij In Pi * ln P j + Yib ln LBlb + ɪ Ylblb (ln LB )
i ij
+ ∑ Yγr ln LB * ln P + δ ln T + 1 δ (ln T)2 + ∑ δ lnT * ln P + δι ln T * ln LB
LBi i T TT Ti i TLB
i 2i
+ δ ln S + 1 δss1 ln(5)2 + ∑ δ ln S *ln P + δsbB ln 5 *ln LB + ξ
S SSi Si i SLB
2i
Greater efficiency can be obtained by estimating the cost share equations jointly with the
cost function. Share equations are given by the derivative of the cost function with respect to
input prices, as expressed in equation 3.