Estimation followed several common practices. First, symmetry and homogeneity of degree one
were imposed on the model. Second, to simplify interpretation, all variables were normalized by
dividing by their sample means. The first order input price terms, βi , could then be interpreted
as the estimated cost share at the sample means of the right-hand side variables. These cost
shares vary as the right-hand side variables depart from their sample means. Notice also that the
model regresses costs in current dollars on prices in current dollars; thus, there is no need to use
deflators or other means to account for inter-temporal price variations due to inflation.
The system of equations includes the cost function and the cost share equations. Since
costs shares sum to one, the capital share equation was dropped to avoid a singular covariance
matrix (the coefficients of one equation, capital in this case, are implied by the other two
equations). Finally, to take account of likely cross equation correlation and to achieve efficiency
gains, the entire system of equations, including the cost function and the cost share equations,
was estimated as a system with a nonlinear, iterative, seemingly unrelated regression procedure.
A four factor cost function that included separate entries for meat and materials was
tested initially. However, the model failed monotonicity tests and was dropped. It was
subsequently determined that the problem lie in the meat and materials data. The identity of total
value of materials equal to the value of animal/meat plus material inputs is supposed to hold and
did for analyses by MacDonald and Ollinger (2000, 2005) and Ollinger, MacDonald, and
Madison (2005). However, the identity failed in the 2002 data. The poor data caused
monotonicity tests to fail and required the use of three factor cost function with one variable (the
price for total materials) equal the meat/liveweight animal input costs plus material costs divided
by the weight of total meat/liveweight animal inputs, as defined earlier. Since materials are a
10